remove on_stack from graph_rewrite

remove restrictions on range ending in indexing (#12543 )
* remove restrictions on range ending in indexing * early simplify * Revert "early simplify" This reverts commit 657d9972c2. * disable const folding tests
2025-10-09 14:14:25 +08:00 · 2025-10-09 13:53:08 +08:00 · 2025-10-09 01:52:30 -04:00 · 2025-10-09 07:14:27 +03:00 · 2025-10-09 06:56:10 +03:00 · 2025-10-09 11:32:34 +08:00
398 changed files with 22032 additions and 229116 deletions
@@ -121,7 +121,7 @@ runs:
        echo 'Acquire::GzipIndexes "true";' | sudo tee /etc/apt/apt.conf.d/gzip
        echo 'Acquire::http::Pipeline-Depth "5";' | sudo tee -a /etc/apt/apt.conf.d/99parallel
        echo 'Binary::apt::APT::Keep-Downloaded-Packages "true";' | sudo tee -a /etc/apt/apt.conf.d/99keep-debs
-    
+
    - name: Add OpenCL Repo
      if: inputs.opencl == 'true' && runner.os == 'Linux'
      shell: bash
@@ -174,7 +174,7 @@ runs:
        if [[ "${{ inputs.llvm }}" == "true" ]]; then
          pkgs+=" libllvm20 clang-20 lld-20"
        fi
-        
+
        echo "pkgs=$pkgs" >> "$GITHUB_OUTPUT"
        echo "hash=$(echo -n "$pkgs" | sha256sum | cut -d' ' -f1)" >> "$GITHUB_OUTPUT"

@@ -183,21 +183,21 @@ runs:
      uses: actions/cache@v4
      with:
        path: /var/cache/apt/archives/
-        key: ${{ runner.os }}-apt-${{ steps.apt-pkgs.outputs.hash }}
+        key: ${{ runner.os }}-apt-${{ steps.apt-pkgs.outputs.hash }}-${{ env.APT_CACHE_VERSION }}

    - name: Run apt Update + Install
      if: runner.os == 'Linux' && (inputs.opencl == 'true' || inputs.amd == 'true' || inputs.cuda == 'true' || inputs.webgpu == 'true' || inputs.llvm == 'true')
      shell: bash
      run: |
        sudo apt -qq update || true
-        
+
        # ******** do install ********
        if [[ -n "${{ steps.apt-pkgs.outputs.pkgs }}" ]]; then
          sudo apt-get -y --allow-unauthenticated --no-install-recommends install ${{ steps.apt-pkgs.outputs.pkgs }}
        fi
-        
+
        sudo chown -R $USER:$USER /var/cache/apt/archives/
-    
+
    # **** AMD ****
    - name: Setup AMD (Linux)
      if: inputs.amd == 'true' && runner.os == 'Linux'
@@ -225,16 +225,25 @@ runs:
    - name: Install gpuocelot dependencies (MacOS)
      if: inputs.ocelot == 'true' && runner.os == 'macOS'
      shell: bash
-      run: brew install --quiet cmake ninja llvm@15 zlib glew flex bison boost zstd ncurses
+      run: |
+        pkgs=(cmake ninja llvm@15 zlib glew flex bison [email protected] zstd ncurses)
+        for f in "${pkgs[@]}"; do
+          brew ls --versions "$f" >/dev/null 2>&1 || brew install --quiet "$f"
+        done
+
+        # Fix boost 1.85 for gpuocelot
+        ln -s /opt/homebrew/opt/[email protected] /opt/homebrew/opt/boost || true
+        ln -s /opt/homebrew/opt/boost/lib/libboost_atomic-mt.dylib /opt/homebrew/opt/boost/lib/libboost_atomic.dylib || true
+        ln -s /opt/homebrew/opt/boost/lib/libboost_thread-mt.dylib /opt/homebrew/opt/boost/lib/libboost_thread.dylib || true
    - name: Cache gpuocelot
      if: inputs.ocelot == 'true'
      id: cache-build
      uses: actions/cache@v4
      env:
-        cache-name: cache-gpuocelot-build
+        cache-name: cache-gpuocelot-build-1
      with:
        path: ${{ github.workspace }}/gpuocelot/ocelot
-        key: ${{ runner.os }}-gpuocelot-b16039dc940dc6bc4ea0a98380495769ff35ed99-rebuild-0
+        key: ${{ runner.os }}-gpuocelot-b16039dc940dc6bc4ea0a98380495769ff35ed99-rebuild-${{ env.BUILD_CACHE_VERSION }}
    - name: Clone/compile gpuocelot
      if: inputs.ocelot == 'true' && steps.cache-build.outputs.cache-hit != 'true'
      shell: bash
@@ -244,7 +253,13 @@ runs:
        git checkout b16039dc940dc6bc4ea0a98380495769ff35ed99
        mkdir build
        cd build
-        cmake .. -Wno-dev -G Ninja -DOCELOT_BUILD_TOOLS=OFF -DCMAKE_BUILD_ALWAYS=0 -DBUILD_TESTS_CUDA=OFF -DCMAKE_POLICY_VERSION_MINIMUM=3.5
+
+        CMAKE_ARGS="-Wno-dev -G Ninja -DOCELOT_BUILD_TOOLS=OFF -DCMAKE_BUILD_ALWAYS=0 -DBUILD_TESTS_CUDA=OFF -DCMAKE_POLICY_VERSION_MINIMUM=3.5"
+        if [[ "${{ runner.os }}" == "macOS" ]]; then
+          CMAKE_ARGS="$CMAKE_ARGS -DBoost_INCLUDE_DIR=$(brew --prefix boost)/include -DBoost_LIBRARY_DIR=$(brew --prefix boost)/lib"
+        fi
+
+        cmake .. $CMAKE_ARGS
        ninja
    - name: Install gpuocelot
      if: inputs.ocelot == 'true'
@@ -0,0 +1,91 @@
+name: Autogen
+env:
+  # increment this when downloads substantially change to avoid the internet
+  DOWNLOAD_CACHE_VERSION: '12'
+  PYTHON_CACHE_VERSION: '3'
+  APT_CACHE_VERSION: '1'
+  BUILD_CACHE_VERSION: '1'
+  CAPTURE_PROCESS_REPLAY: 1
+  GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+  PYTHONPATH: ${{ github.workspace }}
+
+on:
+  push:
+    branches:
+      - master
+  pull_request:
+    paths: 
+    - 'tinygrad/runtime/autogen/**/*'
+  workflow_dispatch:
+    paths: 
+    - 'tinygrad/runtime/autogen/**/*'
+
+jobs:
+  autogen:
+    name: Autogen
+    runs-on: ubuntu-24.04
+    timeout-minutes: 15
+    steps:
+    - name: Checkout Code
+      uses: actions/checkout@v4
+    - name: Setup Environment
+      uses: ./.github/actions/setup-tinygrad
+      with:
+        opencl: 'true'
+        amd: 'true'
+        cuda: 'true'
+        webgpu: 'true'
+        llvm: 'true'
+    - name: Install autogen support packages
+      run: sudo apt-get install -y --no-install-recommends llvm-14-dev libclang-14-dev
+    - name: Verify OpenCL autogen
+      run: |
+        cp tinygrad/runtime/autogen/opencl.py /tmp/opencl.py.bak
+        ./autogen_stubs.sh opencl
+        diff /tmp/opencl.py.bak tinygrad/runtime/autogen/opencl.py
+    - name: Verify CUDA autogen
+      run: |
+        cp tinygrad/runtime/autogen/cuda.py /tmp/cuda.py.bak
+        cp tinygrad/runtime/autogen/nv_gpu.py /tmp/nv_gpu.py.bak
+        ./autogen_stubs.sh cuda
+        ./autogen_stubs.sh nv
+        diff /tmp/cuda.py.bak tinygrad/runtime/autogen/cuda.py
+        diff /tmp/nv_gpu.py.bak tinygrad/runtime/autogen/nv_gpu.py
+    - name: Verify AMD autogen
+      run: |
+        cp tinygrad/runtime/autogen/hsa.py /tmp/hsa.py.bak
+        cp tinygrad/runtime/autogen/kfd.py /tmp/kfd.py.bak
+        cp tinygrad/runtime/autogen/comgr.py /tmp/comgr.py.bak
+        cp tinygrad/runtime/autogen/amd_gpu.py /tmp/amd_gpu.py.bak
+        cp tinygrad/runtime/autogen/sqtt.py /tmp/sqtt.py.bak
+        ./autogen_stubs.sh hsa
+        ./autogen_stubs.sh kfd
+        ./autogen_stubs.sh comgr
+        ./autogen_stubs.sh amd
+        ./autogen_stubs.sh sqtt
+        diff /tmp/hsa.py.bak tinygrad/runtime/autogen/hsa.py
+        diff /tmp/kfd.py.bak tinygrad/runtime/autogen/kfd.py
+        diff /tmp/comgr.py.bak tinygrad/runtime/autogen/comgr.py
+        diff /tmp/amd_gpu.py.bak tinygrad/runtime/autogen/amd_gpu.py
+        diff /tmp/sqtt.py.bak tinygrad/runtime/autogen/sqtt.py
+    - name: Verify Linux autogen
+      run: |
+        cp tinygrad/runtime/autogen/libc.py /tmp/libc.py.bak
+        cp tinygrad/runtime/autogen/io_uring.py /tmp/io_uring.py.bak
+        cp tinygrad/runtime/autogen/ib.py /tmp/ib.py.bak
+        ./autogen_stubs.sh libc
+        ./autogen_stubs.sh io_uring
+        ./autogen_stubs.sh ib
+        diff /tmp/libc.py.bak tinygrad/runtime/autogen/libc.py
+        diff /tmp/io_uring.py.bak tinygrad/runtime/autogen/io_uring.py
+        diff /tmp/ib.py.bak tinygrad/runtime/autogen/ib.py
+    - name: Verify WebGPU autogen
+      run: |
+        cp tinygrad/runtime/autogen/webgpu.py /tmp/webgpu.py.bak
+        ./autogen_stubs.sh webgpu
+        diff /tmp/webgpu.py.bak tinygrad/runtime/autogen/webgpu.py
+    - name: Verify LLVM autogen
+      run: |
+        cp tinygrad/runtime/autogen/llvm.py /tmp/llvm.py.bak
+        ./autogen_stubs.sh llvm
+        diff /tmp/llvm.py.bak tinygrad/runtime/autogen/llvm.py
@@ -28,7 +28,7 @@ jobs:
      # since sudo is required for usbgpu on macos, move the cache to a new location, as some of the files are owned by root
      PYTHONPYCACHEPREFIX: /tmp/tiny_python_pycache
    runs-on: [self-hosted, macOS]
-    timeout-minutes: 20
+    timeout-minutes: 60
    defaults:
      run:
        shell: bash -e -o pipefail {0}
@@ -52,26 +52,28 @@ jobs:
    - name: reset process replay
      run: python3.11 test/external/process_replay/reset.py
    - name: Run Stable Diffusion
-      run: BENCHMARK_LOG=stable_diffusion JIT=1 python3.11 examples/stable_diffusion.py --fp16 --seed 0 --noshow --timing | tee sd.txt
+      run: BENCHMARK_LOG=stable_diffusion JIT=1 ASSERT_MIN_STEP_TIME=800 python3.11 examples/stable_diffusion.py --fp16 --seed 0 --noshow --timing | tee sd.txt
    - name: Run Stable Diffusion without fp16
-      run: BENCHMARK_LOG=stable_diffusion_fp32 JIT=1 python3.11 examples/stable_diffusion.py --seed 0 --noshow --timing | tee sd_no_fp16.txt
+      run: BENCHMARK_LOG=stable_diffusion_fp32 JIT=1 ASSERT_MIN_STEP_TIME=900 python3.11 examples/stable_diffusion.py --seed 0 --noshow --timing | tee sd_no_fp16.txt
    - name: Run Stable Diffusion v2
-      run: BENCHMARK_LOG=stable_diffusion_v2 JIT=1 python3.11 examples/sdv2.py --fp16 --seed 0 --noshow --timing | tee sdv2.txt
+      # TODO: very slow step time
+      run: BENCHMARK_LOG=stable_diffusion_v2 JIT=1 ASSERT_MIN_STEP_TIME=10000 python3.11 examples/sdv2.py --fp16 --seed 0 --noshow --timing | tee sdv2.txt
    # process replay can't capture this, the graph is too large
-    - name: Run SDXL
-      run: BENCHMARK_LOG=stable_diffusion_xl CAPTURE_PROCESS_REPLAY=0 JIT=1 python3.11 examples/sdxl.py --seed 0 --noshow --timing | tee sdxl.txt
+    # TODO: too slow
+    # - name: Run SDXL
+    #   run: BENCHMARK_LOG=stable_diffusion_xl ASSERT_MIN_STEP_TIME=5000 CAPTURE_PROCESS_REPLAY=0 JIT=1 python3.11 examples/sdxl.py --seed 0 --noshow --timing | tee sdxl.txt
    - name: Run model inference benchmark
      run: METAL=1 python3.11 test/external/external_model_benchmark.py
-    - name: Run huggingface_onnx test
-      run: METAL=1 python3.11 extra/huggingface_onnx/run_models.py test --debug FacebookAI/xlm-roberta-large
    - name: Test speed vs torch
-      run: BIG=2 MPS=1 python3.11 test/test_speed_v_torch.py | tee torch_speed.txt
+      run: BIG=2 MPS=1 python3.11 test/speed/external_test_speed_v_torch.py | tee torch_speed.txt
    - name: Test tensor cores
-      run: METAL=1 python3.11 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded TestLinearizer.test_tensor_cores_padded_uops
+      run: METAL=1 python3.11 test/opt/test_tensor_cores.py
    - name: Test AMX tensor cores
      run: |
-        DEBUG=2 CPU=1 AMX=1 python3.11 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded TestLinearizer.test_tensor_cores_padded_uops TestFloat4.test_float4_multidim_amx TestFloat4.test_float4_multidim_unaligned_load_amx
-        DEBUG=2 LLVM=1 AMX=1 python3.11 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded TestLinearizer.test_tensor_cores_padded_uops TestFloat4.test_float4_multidim_amx TestFloat4.test_float4_multidim_unaligned_load_amx
+        DEBUG=2 CPU=1 CPU_LLVM=0 AMX=1 python3.11 test/opt/test_tensor_cores.py
+        DEBUG=2 CPU=1 CPU_LLVM=1 AMX=1 python3.11 test/opt/test_tensor_cores.py
+        DEBUG=2 CPU=1 CPU_LLVM=0 AMX=1 python3.11 test/opt/test_gen_float4.py TestFloat4.test_float4_multidim_amx TestFloat4.test_float4_multidim_unaligned_load_amx
+        DEBUG=2 CPU=1 CPU_LLVM=1 AMX=1 python3.11 test/opt/test_gen_float4.py TestFloat4.test_float4_multidim_amx TestFloat4.test_float4_multidim_unaligned_load_amx
    - name: Run Tensor Core GEMM (float)
      run: DEBUG=2 SHOULD_USE_TC=1 python3.11 extra/gemm/simple_matmul.py | tee matmul.txt
    - name: Run Tensor Core GEMM (half)
@@ -99,7 +101,7 @@ jobs:
    - name: Run GPT2
      run: |
        BENCHMARK_LOG=gpt2_nojit JIT=0 python3.11 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
-        BENCHMARK_LOG=gpt2 JIT=1 python3.11 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
+        BENCHMARK_LOG=gpt2 JIT=1 ASSERT_MIN_STEP_TIME=13 python3.11 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
    - name: Run GPT2 w HALF
      run: BENCHMARK_LOG=gpt2_half HALF=1 python3.11 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half.txt
    - name: Run GPT2 w HALF/BEAM
@@ -108,22 +110,27 @@ jobs:
      run: BENCHMARK_LOG=olmoe python3.11 examples/olmoe.py
    - name: Train MNIST
      run: time PYTHONPATH=. TARGET_EVAL_ACC_PCT=96.0 python3.11 examples/beautiful_mnist.py | tee beautiful_mnist.txt
-    - name: Run 10 CIFAR training steps
-      run: BENCHMARK_LOG=cifar_10steps JIT=1 STEPS=10 python3.11 examples/hlb_cifar10.py | tee train_cifar.txt
-    - name: Run 10 CIFAR training steps w HALF
-      run: BENCHMARK_LOG=cifar_10steps_half JIT=2 STEPS=10 DEFAULT_FLOAT=HALF python3.11 examples/hlb_cifar10.py | tee train_cifar_half.txt
+
+    # NOTE: this is failing in CI. it is not failing on my machine and I don't really have a way to debug it
+    # the error is "RuntimeError: Internal Error (0000000e:Internal Error)"
+    #- name: Run 10 CIFAR training steps
+    #  run: BENCHMARK_LOG=cifar_10steps JIT=1 ASSERT_MIN_STEP_TIME=3000 STEPS=10 python3.11 examples/hlb_cifar10.py | tee train_cifar.txt
+    #- name: Run 10 CIFAR training steps w HALF
+    #  run: BENCHMARK_LOG=cifar_10steps_half JIT=2 ASSERT_MIN_STEP_TIME=3000 STEPS=10 DEFAULT_FLOAT=HALF python3.11 examples/hlb_cifar10.py | tee train_cifar_half.txt
+
    #- name: Run 10 CIFAR training steps w BF16
    #  run: STEPS=10 DEFAULT_FLOAT=BFLOAT16 python3.11 examples/hlb_cifar10.py | tee train_cifar_bf16.txt
-    - name: Run 10 CIFAR training steps w winograd
-      run: BENCHMARK_LOG=cifar_10steps_wino JIT=1 WINO=1 STEPS=10 python3.11 examples/hlb_cifar10.py | tee train_cifar_wino.txt
+    # TODO: too slow
+    # - name: Run 10 CIFAR training steps w winograd
+    #   run: BENCHMARK_LOG=cifar_10steps_wino JIT=1 ASSERT_MIN_STEP_TIME=150 WINO=1 STEPS=10 python3.11 examples/hlb_cifar10.py | tee train_cifar_wino.txt
    - name: UsbGPU boot time
      run: sudo -E PYTHONPATH=. DEBUG=2 AM_RESET=1 AMD=1 AMD_IFACE=USB time python3.11 test/test_tiny.py TestTiny.test_plus
    - name: UsbGPU tiny tests
      run: sudo -E PYTHONPATH=. AMD=1 AMD_IFACE=USB python3.11 test/test_tiny.py
    - name: UsbGPU copy speeds
      run: sudo -E PYTHONPATH=. AMD=1 AMD_IFACE=USB python3.11 test/external/external_test_usb_asm24.py TestDevCopySpeeds
-    - name: UsbGPU openpilot test
-      run: sudo -E PYTHONPATH=. AMD=1 AMD_IFACE=USB NOLOCALS=0 IMAGE=0 GRAPH_ONE_KERNEL=1 python3.11 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/9118973ed03c1ae1d40cf69a29507ec2cc78efd7/selfdrive/modeld/models/supercombo.onnx
+    #- name: UsbGPU openpilot test
+    #  run: sudo -E PYTHONPATH=. AMD=1 AMD_IFACE=USB NOLOCALS=0 IMAGE=0 GRAPH_ONE_KERNEL=1 python3.11 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/9118973ed03c1ae1d40cf69a29507ec2cc78efd7/selfdrive/modeld/models/supercombo.onnx
    - uses: actions/upload-artifact@v4
      with:
        name: Speed (Mac)
@@ -160,7 +167,7 @@ jobs:
  testnvidiabenchmark:
    name: tinybox green Benchmark
    runs-on: [self-hosted, Linux, tinyboxgreen]
-    timeout-minutes: 30
+    timeout-minutes: 60
    defaults:
      run:
        shell: bash -e -o pipefail {0}
@@ -189,22 +196,22 @@ jobs:
    - name: Run model inference benchmark
      run: NV=1 CAPTURE_PROCESS_REPLAY=0 NOCLANG=1 python3 test/external/external_model_benchmark.py
    - name: Test speed vs torch
-      run: NV=1 CAPTURE_PROCESS_REPLAY=0 HALF=1 BIG=2 TORCHCUDA=1 python3 test/test_speed_v_torch.py | tee torch_speed.txt
+      run: NV=1 CAPTURE_PROCESS_REPLAY=0 HALF=1 BIG=2 TORCHCUDA=1 python3 test/speed/external_test_speed_v_torch.py | tee torch_speed.txt
    - name: Test speed vs theoretical
      run: NV=1 IGNORE_BEAM_CACHE=1 BEAM_DEBUG=1 DEBUG=1 python -m pytest -rA test/external/speed_v_theoretical.py --durations=20
    - name: Test benchmark allreduce
      run: NV=1 python test/external/external_benchmark_multitensor_allreduce.py
    - name: Test tensor cores
      run: |
-        NV=1 ALLOW_TF32=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded TestLinearizer.test_tensor_cores_padded_uops
-        PTX=1 ALLOW_TF32=1 NV=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded TestLinearizer.test_tensor_cores_padded_uops
+        NV=1 ALLOW_TF32=1 python3 test/opt/test_tensor_cores.py
+        NV=1 NV_PTX=1 ALLOW_TF32=1 python3 test/opt/test_tensor_cores.py
    - name: Run Tensor Core GEMM (CUDA)
      run: |
        CUDA=1 SHOULD_USE_TC=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul.txt
        CUDA=1 SHOULD_USE_TC=1 BFLOAT16=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul_bfloat16.txt
        CUDA=1 SHOULD_USE_TC=1 ALLOW_TF32=1 DEBUG=2 ATOL=2e-2 python3 extra/gemm/simple_matmul.py | tee matmul_tf32.txt
    - name: Run Tensor Core GEMM (PTX)
-      run: NV=1 PTX=1 SHOULD_USE_TC=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul_ptx.txt
+      run: NV=1 NV_PTX=1 SHOULD_USE_TC=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul_ptx.txt
    - name: Run Tensor Core GEMM (NV)
      run: NV=1 SHOULD_USE_TC=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul_nv.txt
    - name: Test NV=1
@@ -213,8 +220,9 @@ jobs:
      run: DEBUG=2 CUDA=1 python -m pytest -rA test/test_tiny.py
    - name: Run Stable Diffusion
      run: BENCHMARK_LOG=stable_diffusion NV=1 python3 examples/stable_diffusion.py --fp16 --seed 0 --noshow --timing | tee sd.txt
-    - name: Run SDXL
-      run: BENCHMARK_LOG=stable_diffusion_xl CAPTURE_PROCESS_REPLAY=0 NV=1 CAPTURE_PROCESS_REPLAY=0 python3 examples/sdxl.py --seed 0 --noshow --timing | tee sdxl.txt
+    # TODO: too slow
+    # - name: Run SDXL
+    #   run: BENCHMARK_LOG=stable_diffusion_xl ASSERT_MIN_STEP_TIME=2000 CAPTURE_PROCESS_REPLAY=0 NV=1 CAPTURE_PROCESS_REPLAY=0 python3 examples/sdxl.py --seed 0 --noshow --timing | tee sdxl.txt
    - name: Run LLaMA
      run: |
        BENCHMARK_LOG=llama_nojit NV=1 JIT=0 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_unjitted.txt
@@ -238,9 +246,9 @@ jobs:
    - name: Run GPT2
      run: |
        BENCHMARK_LOG=gpt2_nojit NV=1 JIT=0 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
-        BENCHMARK_LOG=gpt2 NV=1 JIT=1 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
+        BENCHMARK_LOG=gpt2 NV=1 JIT=1 ASSERT_MIN_STEP_TIME=4 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
    - name: Run GPT2 w HALF
-      run: BENCHMARK_LOG=gpt2_half NV=1 HALF=1 python3 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half.txt
+      run: BENCHMARK_LOG=gpt2_half NV=1 HALF=1 ASSERT_MIN_STEP_TIME=6 python3 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half.txt
    - name: Run GPT2 w HALF/BEAM
      run: BENCHMARK_LOG=gpt2_half_beam NV=1 HALF=1 JITBEAM=2 IGNORE_BEAM_CACHE=1 python3 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half_beam.txt
    - uses: actions/upload-artifact@v4
@@ -274,7 +282,7 @@ jobs:
  testmorenvidiabenchmark:
    name: tinybox green Training Benchmark
    runs-on: [self-hosted, Linux, tinyboxgreen]
-    timeout-minutes: 20
+    timeout-minutes: 60
    defaults:
      run:
        shell: bash -e -o pipefail {0}
@@ -299,30 +307,33 @@ jobs:
        rm -f /tmp/staging.db /tmp/staging.db-shm /tmp/staging.db-wal
    - name: reset process replay
      run: test/external/process_replay/reset.py
-    - name: Fuzz Padded Tensor Core GEMM (NV)
-      run: NV=1 M_START=12 M_STOP=20 M_STEP=1 N_START=6 N_STOP=10 N_STEP=1 K_START=28 K_STOP=36 K_STEP=1 HALF=1 TC_OPT=2 python3 ./extra/gemm/fuzz_matmul.py
-    - name: Fuzz Padded Tensor Core GEMM (PTX)
-      run: NV=1 PTX=1 M_START=12 M_STOP=20 M_STEP=1 N_START=6 N_STOP=10 N_STEP=1 K_START=28 K_STOP=36 K_STEP=1 HALF=1 TC_OPT=2 python3 ./extra/gemm/fuzz_matmul.py
+    # TODO: too slow
+    # - name: Fuzz Padded Tensor Core GEMM (NV)
+    #   run: NV=1 M_START=12 M_STOP=20 M_STEP=1 N_START=6 N_STOP=10 N_STEP=1 K_START=28 K_STOP=36 K_STEP=1 HALF=1 TC_OPT=2 python3 ./extra/gemm/fuzz_matmul.py
+    # TODO: too slow
+    # - name: Fuzz Padded Tensor Core GEMM (PTX)
+    #   run: NV=1 NV_PTX=1 M_START=12 M_STOP=20 M_STEP=1 N_START=6 N_STOP=10 N_STEP=1 K_START=28 K_STOP=36 K_STEP=1 HALF=1 TC_OPT=2 python3 ./extra/gemm/fuzz_matmul.py
    - name: Train MNIST
      run: time PYTHONPATH=. NV=1 TARGET_EVAL_ACC_PCT=96.0 python3 examples/beautiful_mnist.py | tee beautiful_mnist.txt
    - name: Run 10 CIFAR training steps
-      run: BENCHMARK_LOG=cifar_10steps NV=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar.txt
+      run: BENCHMARK_LOG=cifar_10steps ASSERT_MIN_STEP_TIME=270 NV=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar.txt
    - name: Run 10 CIFAR training steps w HALF
-      run: BENCHMARK_LOG=cifar_10steps_half NV=1 STEPS=10 DEFAULT_FLOAT=HALF python3 examples/hlb_cifar10.py | tee train_cifar_half.txt
+      run: BENCHMARK_LOG=cifar_10steps_half ASSERT_MIN_STEP_TIME=310 NV=1 STEPS=10 DEFAULT_FLOAT=HALF python3 examples/hlb_cifar10.py | tee train_cifar_half.txt
    - name: Run 10 CIFAR training steps w BF16
-      run: BENCHMARK_LOG=cifar_10steps_bf16 NV=1 STEPS=10 DEFAULT_FLOAT=BFLOAT16 python3 examples/hlb_cifar10.py | tee train_cifar_bf16.txt
-    - name: Run 10 CIFAR training steps w winograd
-      run: BENCHMARK_LOG=cifar_10steps_half_wino NV=1 CAPTURE_PROCESS_REPLAY=0 WINO=1 STEPS=10 DEFAULT_FLOAT=HALF python3 examples/hlb_cifar10.py | tee train_cifar_wino.txt
+      run: BENCHMARK_LOG=cifar_10steps_bf16 ASSERT_MIN_STEP_TIME=310 NV=1 STEPS=10 DEFAULT_FLOAT=BFLOAT16 python3 examples/hlb_cifar10.py | tee train_cifar_bf16.txt
+    # TODO: too slow
+    # - name: Run 10 CIFAR training steps w winograd
+    #   run: BENCHMARK_LOG=cifar_10steps_half_wino ASSERT_MIN_STEP_TIME=350 NV=1 CAPTURE_PROCESS_REPLAY=0 WINO=1 STEPS=10 DEFAULT_FLOAT=HALF python3 examples/hlb_cifar10.py | tee train_cifar_wino.txt
    - name: Run full CIFAR training w 1 GPU
-      run: time BENCHMARK_LOG=cifar NV=1 DEFAULT_FLOAT=HALF LATEWINO=1 STEPS=1000 TARGET_EVAL_ACC_PCT=93.2 python3 examples/hlb_cifar10.py | tee train_cifar_one_gpu.txt
+      run: time BENCHMARK_LOG=cifar NV=1 DEFAULT_FLOAT=HALF STEPS=1000 TARGET_EVAL_ACC_PCT=93.0 python3 examples/hlb_cifar10.py | tee train_cifar_one_gpu.txt
    - name: Run full CIFAR training steps w 6 GPUS
-      run: time BENCHMARK_LOG=cifar_6gpu CAPTURE_PROCESS_REPLAY=0 NV=1 DEFAULT_FLOAT=HALF STEPS=350 BS=1536 GPUS=6 TARGET_EVAL_ACC_PCT=93.2 python3 examples/hlb_cifar10.py | tee train_cifar_six_gpu.txt
+      run: time BENCHMARK_LOG=cifar_6gpu CAPTURE_PROCESS_REPLAY=0 NV=1 DEFAULT_FLOAT=HALF STEPS=350 BS=1536 GPUS=6 TARGET_EVAL_ACC_PCT=93.0 python3 examples/hlb_cifar10.py | tee train_cifar_six_gpu.txt
    - name: Run MLPerf resnet eval on training data
      run: time BENCHMARK_LOG=resnet_eval NV=1 MODEL=resnet python3 examples/mlperf/model_eval.py
-    - name: Run 10 MLPerf ResNet50 training steps (1 gpu)
-      run: BENCHMARK_LOG=resnet_10steps NV=1 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=256 GPUS=1 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet_one_gpu.txt
-    - name: Run 10 MLPerf ResNet50 training steps (6 gpu)
-      run: BENCHMARK_LOG=resnet_10steps_6gpu NV=1 CAPTURE_PROCESS_REPLAY=0 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=1536 GPUS=6 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet.txt
+    #- name: Run 10 MLPerf ResNet50 training steps (1 gpu)
+    #  run: BENCHMARK_LOG=resnet_10steps NV=1 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=256 GPUS=1 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet_one_gpu.txt
+    #- name: Run 10 MLPerf ResNet50 training steps (6 gpu)
+    #  run: BENCHMARK_LOG=resnet_10steps_6gpu NV=1 CAPTURE_PROCESS_REPLAY=0 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=1536 GPUS=6 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet.txt
    - name: Run 10 MLPerf Bert training steps (6 gpu)
      # TODO: remove BERT_LAYERS once scheduler is fast
      run: BENCHMARK_LOG=bert_10steps_6gpu NV=1 CAPTURE_PROCESS_REPLAY=0 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=66 GPUS=6 BERT_LAYERS=2 MODEL=bert python3 examples/mlperf/model_train.py | tee train_bert.txt
@@ -346,7 +357,7 @@ jobs:
  testamdbenchmark:
    name: tinybox red Benchmark
    runs-on: [self-hosted, Linux, tinybox]
-    timeout-minutes: 20
+    timeout-minutes: 60
    defaults:
      run:
        shell: bash -e -o pipefail {0}
@@ -391,13 +402,13 @@ jobs:
    #- name: Test speed vs torch
    #  run: |
    #    python3 -c "import torch; print(torch.__version__)"
-    #    LD_PRELOAD="/opt/rocm/lib/libhsa-runtime64.so" HSA=1 BIG=2 TORCHCUDA=1 python3 test/test_speed_v_torch.py | tee torch_speed.txt
+    #    LD_PRELOAD="/opt/rocm/lib/libhsa-runtime64.so" HSA=1 BIG=2 TORCHCUDA=1 python3 test/speed/external_test_speed_v_torch.py | tee torch_speed.txt
    - name: Test speed vs theoretical
      run: AMD=1 IGNORE_BEAM_CACHE=1 BEAM_DEBUG=1 DEBUG=1 python -m pytest -rA test/external/speed_v_theoretical.py --durations=20
    - name: Test tensor cores
      run: |
-        AMD=1 AMD_LLVM=0 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded_amd TestLinearizer.test_tensor_cores_padded_uops
-        AMD=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded_amd TestLinearizer.test_tensor_cores_padded_uops
+        AMD=1 AMD_LLVM=0 python3 test/opt/test_tensor_cores.py
+        AMD=1 AMD_LLVM=1 python3 test/opt/test_tensor_cores.py
        AMD=1 SHOULD_USE_TC=1 BFLOAT16=1 DEBUG=2 python3 extra/gemm/simple_matmul.py
    - name: Run Tensor Core GEMM (AMD)
      run: AMD=1 SHOULD_USE_TC=1 HALF=1 DEBUG=2 ATOL=2e-2 python3 extra/gemm/simple_matmul.py | tee matmul_amd.txt
@@ -415,9 +426,10 @@ jobs:
    - name: Test AM warm start time
      run: time AMD=1 python3 test/test_tiny.py TestTiny.test_plus
    - name: Run Stable Diffusion
-      run: BENCHMARK_LOG=stable_diffusion AMD=1 python3 examples/stable_diffusion.py --fp16 --seed 0 --noshow --timing | tee sd.txt
-    - name: Run SDXL
-      run: BENCHMARK_LOG=stable_diffusion_xl CAPTURE_PROCESS_REPLAY=0 AMD=1 python3 examples/sdxl.py --seed 0 --noshow --timing | tee sdxl.txt
+      run: BENCHMARK_LOG=stable_diffusion ASSERT_MIN_STEP_TIME=550 AMD=1 python3 examples/stable_diffusion.py --fp16 --seed 0 --noshow --timing | tee sd.txt
+    # TODO: too slow
+    # - name: Run SDXL
+    #   run: BENCHMARK_LOG=stable_diffusion_xl ASSERT_MIN_STEP_TIME=3200 CAPTURE_PROCESS_REPLAY=0 AMD=1 python3 examples/sdxl.py --seed 0 --noshow --timing | tee sdxl.txt
    - name: Run LLaMA 7B
      run: |
        BENCHMARK_LOG=llama_nojit AMD=1 JIT=0 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_unjitted.txt
@@ -443,9 +455,9 @@ jobs:
    - name: Run GPT2
      run: |
        BENCHMARK_LOG=gpt2_nojit AMD=1 JIT=0 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
-        BENCHMARK_LOG=gpt2 AMD=1 JIT=1 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
+        BENCHMARK_LOG=gpt2 AMD=1 JIT=1 ASSERT_MIN_STEP_TIME=5 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
    - name: Run GPT2 w HALF
-      run: BENCHMARK_LOG=gpt2_half AMD=1 HALF=1 python3 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half.txt
+      run: BENCHMARK_LOG=gpt2_half AMD=1 HALF=1 ASSERT_MIN_STEP_TIME=5 python3 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half.txt
    - name: Run GPT2 w HALF/BEAM
      run: BENCHMARK_LOG=gpt2_half_beam AMD=1 HALF=1 JITBEAM=2 IGNORE_BEAM_CACHE=1 python3 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half_beam.txt
    - uses: actions/upload-artifact@v4
@@ -476,7 +488,7 @@ jobs:
  testmoreamdbenchmark:
    name: tinybox red Training Benchmark
    runs-on: [self-hosted, Linux, tinybox]
-    timeout-minutes: 30
+    timeout-minutes: 60
    defaults:
      run:
        shell: bash -e -o pipefail {0}
@@ -508,19 +520,20 @@ jobs:
    - name: Train MNIST
      run: time PYTHONPATH=. AMD=1 TARGET_EVAL_ACC_PCT=96.0 python3 examples/beautiful_mnist.py | tee beautiful_mnist.txt
    - name: Run 10 CIFAR training steps
-      run: BENCHMARK_LOG=cifar_10steps AMD=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar.txt
+      run: BENCHMARK_LOG=cifar_10steps ASSERT_MIN_STEP_TIME=330 AMD=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar.txt
    - name: Run 10 CIFAR training steps w HALF
-      run: BENCHMARK_LOG=cifar_10steps_half AMD=1 STEPS=10 DEFAULT_FLOAT=HALF python3 examples/hlb_cifar10.py | tee train_cifar_half.txt
-    - name: Run 10 CIFAR training steps w BF16
-      run: BENCHMARK_LOG=cifar_10steps_bf16 AMD=1 STEPS=10 DEFAULT_FLOAT=BFLOAT16 python3 examples/hlb_cifar10.py | tee train_cifar_bf16.txt
-    - name: Run 10 CIFAR training steps w winograd
-      run: BENCHMARK_LOG=cifar_10steps_half_wino AMD=1 WINO=1 STEPS=10 DEFAULT_FLOAT=HALF python3 examples/hlb_cifar10.py | tee train_cifar_wino.txt
+      run: BENCHMARK_LOG=cifar_10steps_half ASSERT_MIN_STEP_TIME=330 AMD=1 STEPS=10 DEFAULT_FLOAT=HALF python3 examples/hlb_cifar10.py | tee train_cifar_half.txt
+    # - name: Run 10 CIFAR training steps w BF16
+    #   run: BENCHMARK_LOG=cifar_10steps_bf16 ASSERT_MIN_STEP_TIME=288 AMD=1 STEPS=10 DEFAULT_FLOAT=BFLOAT16 python3 examples/hlb_cifar10.py | tee train_cifar_bf16.txt
+    # TODO: too slow
+    # - name: Run 10 CIFAR training steps w winograd
+    #   run: BENCHMARK_LOG=cifar_10steps_half_wino ASSERT_MIN_STEP_TIME=66 AMD=1 WINO=1 STEPS=10 DEFAULT_FLOAT=HALF python3 examples/hlb_cifar10.py | tee train_cifar_wino.txt
    - name: Run full CIFAR training w 1 GPU
-      run: time BENCHMARK_LOG=cifar AMD=1 DEFAULT_FLOAT=HALF LATEWINO=1 STEPS=1000 TARGET_EVAL_ACC_PCT=93.2 python3 examples/hlb_cifar10.py | tee train_cifar_one_gpu.txt
-    - name: Run full CIFAR training steps w 6 GPUS
-      run: time BENCHMARK_LOG=cifar_6gpu AMD=1 DEFAULT_FLOAT=HALF STEPS=350 BS=1536 GPUS=6 TARGET_EVAL_ACC_PCT=93.2 python3 examples/hlb_cifar10.py | tee train_cifar_six_gpu.txt
-    - name: Run full CIFAR training steps w 6 GPUS (REMOTE)
-      run: time BENCHMARK_LOG=cifar_6gpu_remote REMOTE=1 REMOTEDEV=AMD DEFAULT_FLOAT=HALF STEPS=350 BS=1536 GPUS=6 TARGET_EVAL_ACC_PCT=93.2 python3 examples/hlb_cifar10.py | tee train_cifar_six_gpu_remote.txt
+      run: time BENCHMARK_LOG=cifar AMD=1 DEFAULT_FLOAT=HALF STEPS=1000 TARGET_EVAL_ACC_PCT=93.0 python3 examples/hlb_cifar10.py | tee train_cifar_one_gpu.txt
+    #- name: Run full CIFAR training steps w 6 GPUS
+    #  run: time BENCHMARK_LOG=cifar_6gpu AMD=1 DEFAULT_FLOAT=HALF STEPS=350 BS=1536 GPUS=6 TARGET_EVAL_ACC_PCT=93.0 python3 examples/hlb_cifar10.py | tee train_cifar_six_gpu.txt
+    #- name: Run full CIFAR training steps w 6 GPUS (REMOTE)
+    #  run: time BENCHMARK_LOG=cifar_6gpu_remote REMOTE=1 REMOTEDEV=AMD DEFAULT_FLOAT=HALF STEPS=350 BS=1536 GPUS=6 TARGET_EVAL_ACC_PCT=93.0 python3 examples/hlb_cifar10.py | tee train_cifar_six_gpu_remote.txt
    - uses: actions/upload-artifact@v4
      with:
        name: Speed (AMD Training)
@@ -539,7 +552,7 @@ jobs:
  testmlperfamdbenchmark:
    name: tinybox red MLPerf Benchmark
    runs-on: [self-hosted, Linux, tinybox]
-    timeout-minutes: 30
+    timeout-minutes: 60
    defaults:
      run:
        shell: bash -e -o pipefail {0}
@@ -570,10 +583,10 @@ jobs:
      run: test/external/process_replay/reset.py
    - name: Run MLPerf resnet eval
      run: time BENCHMARK_LOG=resnet_eval AMD=1 MODEL=resnet python3 examples/mlperf/model_eval.py
-    - name: Run 10 MLPerf ResNet50 training steps (1 gpu)
-      run: BENCHMARK_LOG=resnet_10steps AMD=1 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=256 GPUS=1 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet_one_gpu.txt
-    - name: Run 10 MLPerf ResNet50 training steps (6 gpu)
-      run: BENCHMARK_LOG=resnet_10steps_6gpu AMD=1 CAPTURE_PROCESS_REPLAY=0 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=1536 GPUS=6 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet.txt
+    #- name: Run 10 MLPerf ResNet50 training steps (1 gpu)
+    #  run: BENCHMARK_LOG=resnet_10steps AMD=1 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=256 GPUS=1 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet_one_gpu.txt
+    #- name: Run 10 MLPerf ResNet50 training steps (6 gpu)
+    #  run: BENCHMARK_LOG=resnet_10steps_6gpu AMD=1 CAPTURE_PROCESS_REPLAY=0 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=1536 GPUS=6 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet.txt
    - name: Run 10 MLPerf Bert training steps (6 gpu)
      # TODO: remove BERT_LAYERS once scheduler is fast
      run: BENCHMARK_LOG=bert_10steps_6gpu AMD=1 CAPTURE_PROCESS_REPLAY=0 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=66 GPUS=6 BERT_LAYERS=2 MODEL=bert python3 examples/mlperf/model_train.py | tee train_bert.txt
@@ -605,28 +618,28 @@ jobs:
        rm -f /tmp/staging.db /tmp/staging.db-shm /tmp/staging.db-wal
    - name: reset process replay
      run: test/external/process_replay/reset.py
-    - name: validate openpilot 0.9.7
-      run: PYTHONPATH=. FLOAT16=0 IMAGE=2 QCOM=1 taskset -c 4-7 python3 test/external/external_benchmark_openpilot.py https://github.com/commaai/openpilot/raw/v0.9.7/selfdrive/modeld/models/supercombo.onnx | tee openpilot_image_0_9_7.txt
-    - name: benchmark openpilot 0.9.7
-      run: BENCHMARK_LOG=openpilot_0_9_7 PYTHONPATH=. QCOM=1 taskset -c 4-7 python3 test/external/external_benchmark_openpilot.py https://github.com/commaai/openpilot/raw/v0.9.7/selfdrive/modeld/models/supercombo.onnx | tee openpilot_0_9_7.txt
-    - name: benchmark openpilot w IMAGE=2 0.9.7
-      run: BENCHMARK_LOG=openpilot_0_9_7_image PYTHONPATH=. NOLOCALS=1 FLOAT16=1 IMAGE=2 QCOM=1 taskset -c 4-7 python3 test/external/external_benchmark_openpilot.py https://github.com/commaai/openpilot/raw/v0.9.7/selfdrive/modeld/models/supercombo.onnx | tee openpilot_image_0_9_7.txt
+    - name: benchmark openpilot 0.9.9 driving_vision
+      run: BENCHMARK_LOG=openpilot_0_9_9_vision PYTHONPATH=. NOLOCALS=1 FLOAT16=1 IMAGE=2 QCOM=1 taskset -c 4-7 python3 test/external/external_benchmark_openpilot.py https://github.com/commaai/openpilot/raw/v0.9.9/selfdrive/modeld/models/driving_vision.onnx
+    - name: benchmark openpilot 0.9.9 driving_policy
+      run: BENCHMARK_LOG=openpilot_0_9_9_policy PYTHONPATH=. NOLOCALS=1 FLOAT16=1 IMAGE=2 QCOM=1 taskset -c 4-7 python3 test/external/external_benchmark_openpilot.py https://github.com/commaai/openpilot/raw/v0.9.9/selfdrive/modeld/models/driving_policy.onnx
+    - name: benchmark openpilot 0.9.9 dmonitoring
+      run: BENCHMARK_LOG=openpilot_0_9_9_dmonitoring PYTHONPATH=. NOLOCALS=1 FLOAT16=1 IMAGE=2 QCOM=1 taskset -c 4-7 python3 test/external/external_benchmark_openpilot.py https://github.com/commaai/openpilot/raw/v0.9.9/selfdrive/modeld/models/dmonitoring_model.onnx
    - name: openpilot compile3 0.9.9 driving_vision
-      run: PYTHONPATH="." QCOM=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/v0.9.9/selfdrive/modeld/models/driving_vision.onnx
+      run: PYTHONPATH="." ASSERT_MIN_STEP_TIME=22 QCOM=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/v0.9.9/selfdrive/modeld/models/driving_vision.onnx
    - name: openpilot compile3 0.9.9 driving_policy
-      run: PYTHONPATH="." QCOM=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/v0.9.9/selfdrive/modeld/models/driving_policy.onnx
+      run: PYTHONPATH="." ASSERT_MIN_STEP_TIME=7 QCOM=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/v0.9.9/selfdrive/modeld/models/driving_policy.onnx
    - name: openpilot compile3 0.9.9 dmonitoring
-      run: PYTHONPATH="." QCOM=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/v0.9.9/selfdrive/modeld/models/dmonitoring_model.onnx
+      run: PYTHONPATH="." ASSERT_MIN_STEP_TIME=15 QCOM=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/v0.9.9/selfdrive/modeld/models/dmonitoring_model.onnx
    - name: openpilot compile3 Space Lab policy + vision
      run: |
-        PYTHONPATH="." QCOM=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://gitlab.com/commaai/openpilot-lfs.git/gitlab-lfs/objects/22aec22a10ce09384d4a4af2a0bbff08d54af7e0c888503508f356fae4ff0e29
-        PYTHONPATH="." QCOM=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://gitlab.com/commaai/openpilot-lfs.git/gitlab-lfs/objects/c824f68646a3b94f117f01c70dc8316fb466e05fbd42ccdba440b8a8dc86914b
+        PYTHONPATH="." ASSERT_MIN_STEP_TIME=4 QCOM=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://gitlab.com/commaai/openpilot-lfs.git/gitlab-lfs/objects/22aec22a10ce09384d4a4af2a0bbff08d54af7e0c888503508f356fae4ff0e29
+        PYTHONPATH="." ASSERT_MIN_STEP_TIME=26 QCOM=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://gitlab.com/commaai/openpilot-lfs.git/gitlab-lfs/objects/c824f68646a3b94f117f01c70dc8316fb466e05fbd42ccdba440b8a8dc86914b
    - name: benchmark MobileNetV2 on DSP
      run: |
        # generate quantized weights
        ln -s /data/home/tiny/tinygrad/extra/datasets/imagenet extra/datasets/imagenet
        ln -s /data/home/tiny/tinygrad/testsig-*.so .
-        PYTHONPATH=. CC=clang-19 CPU=1 QUANT=1 CNT=0 python3 examples/test_onnx_imagenet.py https://github.com/xamcat/mobcat-samples/raw/refs/heads/master/onnx_runtime/InferencingSample/InferencingSample/mobilenetv2-7.onnx /tmp/model.quant.onnx
+        PYTHONPATH=. CC=clang-19 CPU=1 CPU_LLVM=0 QUANT=1 CNT=0 python3 examples/test_onnx_imagenet.py https://github.com/xamcat/mobcat-samples/raw/refs/heads/master/onnx_runtime/InferencingSample/InferencingSample/mobilenetv2-7.onnx /tmp/model.quant.onnx
        # benchmark on DSP with NOOPT=1, the devectorizer has issues
        PYTHONPATH=. CC=clang-19 DSP=1 DONT_REALIZE_EXPAND=1 NOOPT=1 CNT=2 DEBUG=2 python3 examples/test_onnx_imagenet.py /tmp/model.quant.onnx
    - name: Run process replay tests
@@ -645,7 +658,7 @@ jobs:
  testreddriverbenchmark:
    name: AM Benchmark
    runs-on: [self-hosted, Linux, tinyboxrandom]
-    timeout-minutes: 15
+    timeout-minutes: 20
    defaults:
      run:
        shell: bash -e -o pipefail {0}
@@ -681,8 +694,8 @@ jobs:
    # Fails on 9070
    # - name: Test tensor cores
    #   run: |
-    #     AMD=1 AMD_LLVM=0 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded_amd TestLinearizer.test_tensor_cores_padded_uops
-    #     AMD=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded_amd TestLinearizer.test_tensor_cores_padded_uops
+    #     AMD=1 AMD_LLVM=0 python3 test/test_linearizer.py test/opt/test_tensor_cores.py
+    #     AMD=1 AMD_LLVM=1 python3 test/test_linearizer.py test/opt/test_tensor_cores.py
    #     AMD=1 SHOULD_USE_TC=1 BFLOAT16=1 DEBUG=2 python3 extra/gemm/simple_matmul.py
    - name: Run Tensor Core GEMM (AMD)
      run: AMD=1 SHOULD_USE_TC=1 HALF=1 DEBUG=2 ATOL=2e-2 python3 extra/gemm/simple_matmul.py | tee am_matmul_amd.txt
@@ -690,8 +703,12 @@ jobs:
      run: DEBUG=2 AMD=1 python -m pytest -rA test/test_tiny.py
    - name: Test DISK copy time
      run: AMD=1 TESTFILE=/raid/downloads/llama3-8b-sfr/model-00001-of-00004.safetensors python3 test/external/external_benchmark_disk_raw.py
+    - name: Test CPU copy time
+      run: |
+        AMD=1 GRAPH_ONE_KERNEL=1 PYTHONPATH=. NSZ=8192 python3 test/speed/external_test_copy_speed.py TestCopySpeed.testCopyDefaulttoCPUJit
+        AMD=1 GRAPH_ONE_KERNEL=1 PYTHONPATH=. NSZ=8192 python3 test/speed/external_test_copy_speed.py TestCopySpeed.testCopyCPUtoDefaultJit
    - name: Run full CIFAR training w 1 GPU
-      run: time BENCHMARK_LOG=cifar AMD=1 DEFAULT_FLOAT=HALF LATEWINO=1 STEPS=1000 TARGET_EVAL_ACC_PCT=93.2 python3 examples/hlb_cifar10.py | tee am_train_cifar_one_gpu.txt
+      run: time BENCHMARK_LOG=cifar AMD=1 DEFAULT_FLOAT=HALF STEPS=1000 TARGET_EVAL_ACC_PCT=93.0 python3 examples/hlb_cifar10.py | tee am_train_cifar_one_gpu.txt
    # TODO: enable
    # - name: Run 10 MLPerf ResNet50 training steps (1 gpu)
    #   run: BENCHMARK_LOG=resnet_10steps AMD=1 MNISTMOCK=1 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=256 GPUS=1 MODEL=resnet python3 examples/mlperf/model_train.py | tee am_train_resnet_one_gpu.txt
@@ -712,7 +729,7 @@ jobs:
  testgreendriverbenchmark:
    name: NV Benchmark
    runs-on: [self-hosted, Linux, tinyboxrandom]
-    timeout-minutes: 15
+    timeout-minutes: 20
    defaults:
      run:
        shell: bash -e -o pipefail {0}
@@ -744,15 +761,19 @@ jobs:
    - name: Test driver start time
      run: time DEBUG=3 NV=1 python3 test/test_tiny.py TestTiny.test_plus
    - name: Test tensor cores
-      run: NV=1 ALLOW_TF32=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded TestLinearizer.test_tensor_cores_padded_uops
+      run: NV=1 ALLOW_TF32=1 python3 test/opt/test_tensor_cores.py
    - name: Test DISK copy time
      run: NV=1 TESTFILE=/raid/downloads/llama3-8b-sfr/model-00001-of-00004.safetensors python3 test/external/external_benchmark_disk_raw.py
+    - name: Test CPU copy time
+      run: |
+        NV=1 GRAPH_ONE_KERNEL=1 PYTHONPATH=. NSZ=8192 python3 test/speed/external_test_copy_speed.py TestCopySpeed.testCopyDefaulttoCPUJit
+        NV=1 GRAPH_ONE_KERNEL=1 PYTHONPATH=. NSZ=8192 python3 test/speed/external_test_copy_speed.py TestCopySpeed.testCopyCPUtoDefaultJit
    - name: Test LLAMA-3
      run: BENCHMARK_LOG=llama3_beam NV=1 JITBEAM=2 IGNORE_BEAM_CACHE=1 python3 examples/llama3.py --size 8B --benchmark --temperature 0 | tee nv_llama3_beam.txt
    - name: Run full CIFAR training w 1 GPU
-      run: time BENCHMARK_LOG=cifar NV=1 DEFAULT_FLOAT=HALF LATEWINO=1 STEPS=1000 TARGET_EVAL_ACC_PCT=93.2 python3 examples/hlb_cifar10.py | tee nv_train_cifar_one_gpu.txt
-    - name: Run 10 MLPerf ResNet50 training steps (1 gpu)
-      run: BENCHMARK_LOG=resnet_10steps NV=1 MNISTMOCK=1 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=256 GPUS=1 MODEL=resnet python3 examples/mlperf/model_train.py | tee nv_train_resnet_one_gpu.txt
+      run: time BENCHMARK_LOG=cifar NV=1 DEFAULT_FLOAT=HALF STEPS=1000 TARGET_EVAL_ACC_PCT=93.0 python3 examples/hlb_cifar10.py | tee nv_train_cifar_one_gpu.txt
+    #- name: Run 10 MLPerf ResNet50 training steps (1 gpu)
+    #  run: BENCHMARK_LOG=resnet_10steps NV=1 MNISTMOCK=1 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=256 GPUS=1 MODEL=resnet python3 examples/mlperf/model_train.py | tee nv_train_resnet_one_gpu.txt
    - name: Run 10 MLPerf Bert training steps (1 gpu)
      # TODO: remove BERT_LAYERS once scheduler is fast
      run: BENCHMARK_LOG=bert_10steps NV=1 CAPTURE_PROCESS_REPLAY=0 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=66 GPUS=1 BERT_LAYERS=2 MODEL=bert python3 examples/mlperf/model_train.py | tee nv_train_bert_one_gpu.txt
@@ -20,12 +20,6 @@ repos:
        language: system
        always_run: true
        pass_filenames: false
-      - id: devicetests
-        name: select GPU tests
-        entry: env GPU=1 PYTHONPATH="." python3 -m pytest test/test_uops.py test/test_search.py
-        language: system
-        always_run: true
-        pass_filenames: false
      - id: tests
        name: subset of tests
        entry: env PYTHONPATH="." python3 -m pytest -n=4 test/test_ops.py test/test_dtype.py test/test_schedule.py test/test_assign.py
@@ -54,11 +54,12 @@ confidence=
 # --enable=similarities". If you want to run only the classes checker, but have
 # no Warning level messages displayed, use"--disable=all --enable=classes
 # --disable=W"
-disable=C,R,W0613,W0511,W0212,W0201,W0106,W0603,W0621,W0703,W1201,W1203,E1136,W1514,E1101,W0221,W0105,E0401,abstract-method
+disable=C,R,W0613,W0511,W0212,W0201,W0106,W0603,W0621,W0703,W1201,W1203,E1136,W1514,E1101,W0221,W0105,E0401,abstract-method,W0707
 # E1101 for function binding
 # W0221 for Function class
 # W0105 for comment strings
 # E0401 for missing imports
+# W0707 for not reraising

 # Enable the message, report, category or checker with the given id(s). You can
 # either give multiple identifier separated by comma (,) or put this option
@@ -79,9 +79,8 @@ See [examples/beautiful_mnist.py](examples/beautiful_mnist.py) for the full vers

 tinygrad already supports numerous accelerators, including:

- [x] [GPU (OpenCL)](tinygrad/runtime/ops_gpu.py)
- [x] [CPU (C Code)](tinygrad/runtime/ops_cpu.py)
- [x] [LLVM](tinygrad/runtime/ops_llvm.py)
+- [x] [OpenCL](tinygrad/runtime/ops_cl.py)
+- [x] [CPU](tinygrad/runtime/ops_cpu.py)
 - [x] [METAL](tinygrad/runtime/ops_metal.py)
 - [x] [CUDA](tinygrad/runtime/ops_cuda.py)
 - [x] [AMD](tinygrad/runtime/ops_amd.py)
@@ -198,11 +198,7 @@ generate_amd() {
  clang2py -k cdefstum \
    extra/hip_gpu_driver/sdma_registers.h \
    extra/hip_gpu_driver/nvd.h \
-    extra/hip_gpu_driver/kfd_pm4_headers_ai.h \
-    extra/hip_gpu_driver/soc21_enum.h \
-    extra/hip_gpu_driver/sdma_v6_0_0_pkt_open.h \
    extra/hip_gpu_driver/gc_11_0_0_offset.h \
-    extra/hip_gpu_driver/gc_10_3_0_offset.h \
    extra/hip_gpu_driver/sienna_cichlid_ip_offset.h \
    --clang-args="-I/opt/rocm/include -x c++" \
    -o $BASE/amd_gpu.py
@@ -376,26 +372,6 @@ generate_am() {
    -o $BASE/am/pm4_nv.py
  fixup $BASE/am/pm4_nv.py

-  clang2py -k cdefstum \
-    $AMKERN_INC/vega10_enum.h \
-    -o $BASE/am/vega10.py
-  fixup $BASE/am/vega10.py
-
-  clang2py -k cdefstum \
-    $AMKERN_INC/navi10_enum.h \
-    -o $BASE/am/navi10.py
-  fixup $BASE/am/navi10.py
-
-  clang2py -k cdefstum \
-    $AMKERN_INC/soc21_enum.h \
-    -o $BASE/am/soc21.py
-  fixup $BASE/am/soc21.py
-
-  clang2py -k cdefstum \
-    $AMKERN_INC/soc24_enum.h \
-    -o $BASE/am/soc24.py
-  fixup $BASE/am/soc24.py
-
  clang2py -k cdefstum \
    extra/hip_gpu_driver/sdma_registers.h \
    $AMKERN_AMD/amdgpu/vega10_sdma_pkt_open.h \
@@ -42,7 +42,6 @@ import struct
 from tinygrad.dtype import dtypes
 from tinygrad.device import Buffer, Device
 from tinygrad.uop.ops import UOp, Ops
-from tinygrad.shape.shapetracker import ShapeTracker

 # allocate some buffers + load in values
 out = Buffer(DEVICE, 1, dtypes.int32).allocate()
@@ -51,13 +50,14 @@ b = Buffer(DEVICE, 1, dtypes.int32).allocate().copyin(memoryview(bytearray(struc
 # NOTE: a._buf is the same as the return from cpu.allocator.alloc

 # describe the computation
+idx = UOp.const(dtypes.index, 0)
 buf_1 = UOp(Ops.DEFINE_GLOBAL, dtypes.int32.ptr(), (), 1)
 buf_2 = UOp(Ops.DEFINE_GLOBAL, dtypes.int32.ptr(), (), 2)
-ld_1 = UOp(Ops.LOAD, dtypes.int32, (buf_1.view(ShapeTracker.from_shape((1,))),))
-ld_2 = UOp(Ops.LOAD, dtypes.int32, (buf_2.view(ShapeTracker.from_shape((1,))),))
+ld_1 = UOp(Ops.LOAD, dtypes.int32, (buf_1.index(idx),))
+ld_2 = UOp(Ops.LOAD, dtypes.int32, (buf_2.index(idx),))
 alu = ld_1 + ld_2
 output_buf = UOp(Ops.DEFINE_GLOBAL, dtypes.int32.ptr(), (), 0)
-st_0 = UOp(Ops.STORE, dtypes.void, (output_buf.view(ShapeTracker.from_shape((1,))), alu))
+st_0 = UOp(Ops.STORE, dtypes.void, (output_buf.index(idx), alu))
 s = UOp(Ops.SINK, dtypes.void, (st_0,))

 # convert the computation to a "linearized" format (print the format)
@@ -80,7 +80,8 @@ print("******** third, the UOp ***********")

 from tinygrad.engine.realize import run_schedule
 from tinygrad.engine.schedule import create_schedule_with_vars
-from tinygrad.schedule.kernelize import get_kernelize_map
+from tinygrad.helpers import RANGEIFY
+from tinygrad.schedule.rangeify import get_rangeify_map

 # allocate some values + load in values
 a = UOp.new_buffer(DEVICE, 1, dtypes.int32)
@@ -93,10 +94,10 @@ out = a + b
 s = UOp(Ops.SINK, dtypes.void, (out,))

 # group the computation into kernels
-becomes_map = get_kernelize_map(s)
+becomes_map = get_rangeify_map(s)

 # the compute maps to an assign
-assign = becomes_map[a+b]
+assign = becomes_map[a+b].base

 # the first source is the output buffer (data)
 assert assign.src[0].op is Ops.BUFFER
@@ -10,7 +10,7 @@ Directories are listed in order of how they are processed.

 Group UOps into kernels.

-::: tinygrad.schedule.kernelize.get_kernelize_map
+::: tinygrad.schedule.rangeify.get_rangeify_map
    options:
        members: false
        show_labels: false
@@ -22,12 +22,6 @@ Group UOps into kernels.

 Transforms the ast into an optimized ast. This is where BEAM search and heuristics live.

-::: tinygrad.codegen.opt.get_optimized_ast
-    options:
-        members: false
-        show_labels: false
-        show_source: false
-
 ---

 ## tinygrad/codegen
@@ -3,7 +3,7 @@
 This is a list of environment variable that control the runtime behavior of tinygrad and its examples.
 Most of these are self-explanatory, and are usually used to set an option at runtime.

-Example: `GPU=1 DEBUG=4 python3 -m pytest`
+Example: `CL=1 DEBUG=4 python3 -m pytest`

 However you can also decorate a function to set a value only inside that function.

@@ -31,19 +31,16 @@ These control the behavior of core tinygrad even when used as a library.
 Variable | Possible Value(s) | Description
 ---|---|---
 DEBUG               | [1-7]      | enable debugging output (operations, timings, speed, generated code and more)
-GPU                 | [1]        | enable the GPU (OpenCL) backend
+CL                  | [1]        | enable OpenCL backend
 CUDA                | [1]        | enable CUDA backend
 AMD                 | [1]        | enable AMD backend
 NV                  | [1]        | enable NV backend
 METAL               | [1]        | enable Metal backend (for Mac M1 and after)
-CPU                 | [1]        | enable CPU (Clang) backend
-LLVM                | [1]        | enable LLVM backend
+CPU                 | [1]        | enable CPU backend
 BEAM                | [#]        | number of beams in kernel beam search
 DEFAULT_FLOAT       | [HALF, ...]| specify the default float dtype (FLOAT32, HALF, BFLOAT16, FLOAT64, ...), default to FLOAT32
 IMAGE               | [1-2]      | enable 2d specific optimizations
 FLOAT16             | [1]        | use float16 for images instead of float32
-PTX                 | [1]        | enable the specialized [PTX](https://docs.nvidia.com/cuda/parallel-thread-execution/) assembler for Nvidia GPUs. If not set, defaults to generic CUDA codegen backend.
-PROFILE             | [1]        | enable profiling. This feature is supported in NV, AMD, QCOM and METAL backends.
 VISIBLE_DEVICES     | [list[int]]| restricts the NV/AMD devices that are available. The format is a comma-separated list of identifiers (indexing starts with 0).
 JIT                 | [0-2]      | 0=disabled, 1=[jit enabled](quickstart.md#jit) (default), 2=jit enabled, but graphs are disabled
 VIZ                 | [1]        | 0=disabled, 1=[viz enabled](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/viz)
@@ -2,17 +2,17 @@

 tinygrad supports various runtimes, enabling your code to scale across a wide range of devices. The default runtime can be automatically selected based on the available hardware, or you can force a specific runtime to be default using environment variables (e.g., `CPU=1`).

-| Runtime | Description | Requirements |
-|---------|-------------|--------------|
-| [NV](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_nv.py) | Provides acceleration for NVIDIA GPUs | Ampere/Ada series GPUs |
-| [AMD](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_amd.py) | Provides acceleration for AMD GPUs | RDNA2/RDNA3/RDNA4 series GPUs. You can select one of the interfaces for communication by setting `AMD_IFACE=(KFD|PCI)`. See [AMD interfaces](#amd-interfaces) for more details. |
-| [QCOM](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_qcom.py) | Provides acceleration for QCOM GPUs | 6xx series GPUs |
-| [METAL](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_metal.py) | Utilizes Metal for acceleration on Apple devices | M1+ Macs; Metal 3.0+ for `bfloat` support |
-| [CUDA](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_cuda.py) | Utilizes CUDA for acceleration on NVIDIA GPUs | NVIDIA GPU with CUDA support |
-| [GPU (OpenCL)](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_gpu.py) | Accelerates computations using OpenCL on GPUs | OpenCL 2.0 compatible device |
-| [CPU (C Code)](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_cpu.py) | Runs on CPU using the clang compiler | `clang` compiler in system `PATH` |
-| [LLVM (LLVM IR)](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_llvm.py) | Runs on CPU using the LLVM compiler infrastructure | llvm libraries installed and findable |
-| [WEBGPU](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_webgpu.py) | Runs on GPU using the Dawn WebGPU engine (used in Google Chrome) | Dawn library installed and findable. Download binaries [here](https://github.com/wpmed92/pydawn/releases/tag/v0.3.0). |
+| Runtime | Description | Compiler Options | Requirements |
+|---------|-------------|------------------|--------------|
+| [NV](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_nv.py) | Provides acceleration for NVIDIA GPUs | nvrtc (default)<br>PTX (`NV_PTX=1`) | Ampere/Ada/Blackwell series GPUs.<br>You can select an interface via `NV_IFACE=(NVK\|PCI)`. See [NV interfaces](#nv-interfaces) for details. |
+| [AMD](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_amd.py) | Provides acceleration for AMD GPUs | LLVM (`AMD_LLVM=1`)<br>HIP/COMGR (`AMD_HIP=1`) | RDNA2 or newer GPUs.<br>You can select an interface via `AMD_IFACE=(KFD\|PCI\|USB)`. See [AMD interfaces](#amd-interfaces) for details. |
+| [QCOM](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_qcom.py) | Provides acceleration for QCOM GPUs | - | 6xx series GPUs |
+| [METAL](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_metal.py) | Utilizes Metal for acceleration on Apple devices | - | M1+ Macs; Metal 3.0+ for `bfloat` support |
+| [CUDA](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_cuda.py) | Utilizes CUDA for acceleration on NVIDIA GPUs | nvrtc (default)<br> PTX (`CUDA_PTX=1`) | NVIDIA GPU with CUDA support |
+| [CL](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_cl.py) | Accelerates computations using OpenCL on GPUs | - | OpenCL 2.0 compatible device |
+| [CPU](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_cpu.py) | Runs on CPU using the clang or llvm compiler | Clang JIT (default)<br>LLVM IR (`CPU_LLVM=1`) | `clang` compiler in system `PATH` |
+| [WEBGPU](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_webgpu.py) | Runs on GPU using the Dawn WebGPU engine (used in Google Chrome) | - | Dawn library installed and discoverable. Binaries: [pydawn v0.3.0](https://github.com/wpmed92/pydawn/releases/tag/v0.3.0) |
+

 ## Interoperability

@@ -70,5 +70,12 @@ AMD backend supports several interfaces for communicating with devices:

 * `KFD`: uses the amdgpu driver
 * `PCI`: uses the [AM driver](developer/am.md)
+* `USB`: USB3 interafce for asm24xx chips.

 You can force an interface by setting `AMD_IFACE` to one of these values. In the case of `AMD_IFACE=PCI`, this may unbind your GPU from the amdgpu driver.
+
+## NV Interfaces
+NV backend supports several interfaces for communicating with devices:
+
+* `NVK`: uses the nvidia driver
+* `PCI`: uses the [NV driver](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/support/nv/nvdev.py)
@@ -78,6 +78,7 @@ Elementwise ops operate on a per element basis. They don't change the shape of t
 ::: tinygrad.Tensor.minimum
 ::: tinygrad.Tensor.where
 ::: tinygrad.Tensor.copysign
+::: tinygrad.Tensor.logaddexp

 ## Casting Ops

@@ -6,7 +6,7 @@ If you don't have a tinybox and you want one, see [tinygrad.org](https://tinygra

 ## Welcome

-Welcome to your tinybox! The tinybox is the universal system purpose-built for all AI infrastructure and workloads, from training to inference. The red box includes six 7900XTX GPUs, and the green box includes six 4090 GPUs. Whether you bought a red one or a green one, we want you to love it.
+Welcome to your tinybox! The tinybox is the universal system purpose-built for all AI infrastructure and workloads, from training to inference. The red box includes six 7900XTX GPUs, the green box includes six 4090 GPUs, and the green v2 box includes four 5090 GPUs. Whether you bought a red one or a green one, we want you to love it.

 We don't have a stupid cloud service, you don't have to create a tiny account to set it up, and we aren't tracking how you use the box. We're just happy you bought one. This petaflop is your petaflop.

@@ -2,7 +2,6 @@ import time
 start_tm = time.perf_counter()
 import math
 from typing import Tuple, cast
-import numpy as np
 from tinygrad import Tensor, nn, GlobalCounters, TinyJit, dtypes, Device
 from tinygrad.helpers import partition, trange, getenv, Context
 from extra.lr_scheduler import OneCycleLR
@@ -11,7 +10,7 @@ GPUS = [f'{Device.DEFAULT}:{i}' for i in range(getenv("GPUS", 1))]

 # override tinygrad defaults
 dtypes.default_float = dtypes.half
-Context(FUSE_ARANGE=1, FUSE_OPTIM=1).__enter__()
+Context(FUSE_OPTIM=1).__enter__()

 # from https://github.com/tysam-code/hlb-CIFAR10/blob/main/main.py
 batchsize = getenv("BS", 1024)
@@ -150,13 +149,12 @@ if __name__ == "__main__":
      acc.append((out.argmax(-1) == Y).sum() / eval_batchsize)
    return Tensor.stack(*loss).mean() / (batchsize*loss_batchsize_scaler), Tensor.stack(*acc).mean()

-  np.random.seed(1337)
+  Tensor.manual_seed(1337)
+  num_train_samples = X_train.shape[0]
+
  for epoch in range(math.ceil(hyp['misc']['train_epochs'])):
-    # TODO: move to tinygrad
    gst = time.perf_counter()
-    idxs = np.arange(X_train.shape[0])
-    np.random.shuffle(idxs)
-    tidxs = Tensor(idxs, dtype='int')[:num_steps_per_epoch*batchsize].reshape(num_steps_per_epoch, batchsize)  # NOTE: long doesn't fold
+    tidxs = Tensor.randperm(num_train_samples, dtype='int')[:num_steps_per_epoch*batchsize].reshape(num_steps_per_epoch, batchsize)
    train_loss:float = 0
    for epoch_step in (t:=trange(num_steps_per_epoch)):
      st = time.perf_counter()
@@ -21,7 +21,7 @@ if __name__ == "__main__":
  X_train, Y_train, X_test, Y_test = mnist(fashion=getenv("FASHION"))

  model = Model()
-  opt = nn.optim.Adam(nn.state.get_parameters(model))
+  opt = (nn.optim.Adam if not getenv("MUON") else nn.optim.Muon)(nn.state.get_parameters(model))

  @TinyJit
  @Tensor.train()
@@ -29,8 +29,7 @@ if __name__ == "__main__":
    opt.zero_grad()
    samples = Tensor.randint(getenv("BS", 512), high=X_train.shape[0])
    loss = model(X_train[samples]).sparse_categorical_crossentropy(Y_train[samples]).backward()
-    opt.step()
-    return loss
+    return loss.realize(*opt.schedule_step())

  @TinyJit
  def get_test_acc() -> Tensor: return (model(X_test).argmax(axis=1) == Y_test).mean()*100
@@ -1,6 +1,6 @@
 import sys, time
 from tinygrad import TinyJit, GlobalCounters, fetch, getenv
-from tinygrad.frontend.onnx import OnnxRunner
+from tinygrad.nn.onnx import OnnxRunner
 from extra.onnx_helpers import get_example_inputs, validate

 def load_onnx_model(onnx_file):
@@ -8,7 +8,7 @@ import numpy as np
 import subprocess
 import tensorflow as tf
 import tf2onnx
-from tinygrad.frontend.onnx import OnnxRunner
+from tinygrad.nn.onnx import OnnxRunner
 from tinygrad.tensor import Tensor
 from tinygrad.helpers import to_mv
 from extra.export_model import export_model_clang, compile_net, jit_model
@@ -26,8 +26,8 @@ class Attention:
      start_pos = start_pos.val

    if HALF: x = x.half()
-    xqkv = self.c_attn(x)
-    xq, xk, xv = [xqkv.shrink((None, None, (i*self.dim, (i+1)*self.dim))).reshape(None, None, self.n_heads, self.head_dim) for i in range(3)]
+    xqkv = self.c_attn(x).reshape(None, None, 3, self.n_heads, self.head_dim)
+    xq, xk, xv = [xqkv[:, :, i, :, :] for i in range(3)]
    bsz, seqlen, _, _ = xq.shape

    # create kv cache
@@ -35,11 +35,11 @@ class Attention:
      self.cache_kv = Tensor.zeros(2, bsz, MAX_CONTEXT, self.n_heads, self.head_dim, dtype=x.dtype).contiguous().realize()

    # update the cache
-    self.cache_kv.shrink((None, None,(start_pos,start_pos+seqlen),None,None)).assign(Tensor.stack(xk, xv)).realize()
+    self.cache_kv[:, :, start_pos:start_pos+seqlen, :, :].assign(Tensor.stack(xk, xv)).realize()

    if start_pos > 0:
-      keys = self.cache_kv[0].shrink((None, (0, start_pos+seqlen), None, None))
-      values = self.cache_kv[1].shrink((None, (0, start_pos+seqlen), None, None))
+      keys = self.cache_kv[0][:, :start_pos+seqlen, :, :]
+      values = self.cache_kv[1][:, :start_pos+seqlen, :, :]
    else:
      keys = xk
      values = xv
@@ -64,7 +64,7 @@ class TransformerBlock:

  def __call__(self, x:Tensor, start_pos:Variable, mask:Optional[Tensor]):
    h = x + self.attn(self.ln_1(x), start_pos, mask).float()
-    return (h + self.mlp(self.ln_2(h)))
+    return (h + self.mlp(self.ln_2(h))).contiguous()

 class Transformer:
  def __init__(self, dim, n_heads, n_layers, norm_eps, vocab_size, max_seq_len=1024):
@@ -181,6 +181,7 @@ class GPT2:
    self.tokenizer = tokenizer

  def generate(self, prompt:str, max_length:int, temperature:float, timing:bool=False, batch_size:int=1):
+    step_times = []
    prompt_tokens = self.tokenizer.encode(prompt, allowed_special={"<|endoftext|>"})
    toks = [prompt_tokens[:] for _ in range(batch_size)]
    start_pos = 0
@@ -188,7 +189,7 @@ class GPT2:
      GlobalCounters.reset()
      if timing: print("")
      st = GlobalCounters.time_sum_s
-      with Timing("ran model in ", on_exit=(lambda et: (f", {(GlobalCounters.time_sum_s-st)*1e3:.2f} ms on GPU" if DEBUG>=2 else "")+
+      with Timing("ran model in ", on_exit=(lambda et: (f", {(GlobalCounters.time_sum_s-st)*1e3:.2f} ms on {Device.DEFAULT}" if DEBUG>=2 else "")+
                  f", {GlobalCounters.global_ops*1e-9:.2f} GOPS, {GlobalCounters.global_mem*1e-9:.2f} GB"+
                  (f", {GlobalCounters.global_mem*1e-9/(GlobalCounters.time_sum_s-st):.2f} GB/s" if DEBUG>=2 else "")) if DEBUG else None, enabled=timing):
        with WallTimeEvent(BenchEvent.STEP):
@@ -197,8 +198,13 @@ class GPT2:
          else:
            tokens = Tensor([x[start_pos:] for x in toks])
          tok = self.model(tokens, Variable("start_pos", 1 if start_pos else 0, MAX_CONTEXT-1).bind(start_pos), temperature).tolist()
+      step_times.append((GlobalCounters.time_sum_s-st)*1e3)
      start_pos = len(toks[0])
      for i,t in enumerate(tok): toks[i].append(t)
+
+    if (assert_time:=getenv("ASSERT_MIN_STEP_TIME")):
+      min_time = min(step_times)
+      assert min_time < assert_time, f"Speed regression, expected min step time of < {assert_time} ms but took: {min_time} ms"
    return [self.tokenizer.decode(x) for x in toks]

 # **** main code ****
@@ -118,7 +118,7 @@ class SpeedyResNet:
 # hyper-parameters were exactly the same as the original repo
 bias_scaler = 58
 hyp = {
-  'seed' : 200,
+  'seed' : 201,
  'opt': {
    'bias_lr':            1.76 * bias_scaler/512,
    'non_bias_lr':        1.76 / 512,
@@ -145,7 +145,6 @@ hyp = {
  },
 }

-@Context(FUSE_ARANGE=getenv("FUSE_ARANGE", 1))
 def train_cifar():

  def set_seed(seed):
@@ -229,7 +228,8 @@ def train_cifar():
    if getenv("RANDOM_CROP", 1):
      X = random_crop(X, crop_size=32)
    if getenv("RANDOM_FLIP", 1):
-      X = (Tensor.rand(X.shape[0],1,1,1) < 0.5).where(X.flip(-1), X) # flip LR
+      # NOTE: RANGEIFY=1 needs this contiguous or the X[perms] is very slow
+      X = (Tensor.rand(X.shape[0],1,1,1) < 0.5).where(X.flip(-1), X).contiguous() # flip LR
    X, Y = X[perms], Y[perms]
    return X, Y, *cutmix(X, Y, perms, mask_size=hyp['net']['cutmix_size'])

@@ -355,7 +355,7 @@ def train_cifar():

  # https://www.anandtech.com/show/16727/nvidia-announces-geforce-rtx-3080-ti-3070-ti-upgraded-cards-coming-in-june
  # 136 TFLOPS is the theoretical max w float16 on 3080 Ti
-
+  step_times = []
  model_ema: Optional[modelEMA] = None
  projected_ema_decay_val = hyp['ema']['decay_base'] ** hyp['ema']['every_n_steps']
  i = 0
@@ -413,12 +413,17 @@ def train_cifar():
          model_ema.update(model, Tensor([projected_ema_decay_val*(i/STEPS)**hyp['ema']['decay_pow']]))

      cl = time.monotonic()
+      step_times.append((cl-st)*1000.0)
      device_str = loss.device if isinstance(loss.device, str) else f"{loss.device[0]} * {len(loss.device)}"
      #  53  221.74 ms run,    2.22 ms python,  219.52 ms CL,  803.39 loss, 0.000807 LR, 4.66 GB used,   3042.49 GFLOPS,    674.65 GOPS
      print(f"{i:3d} {(cl-st)*1000.0:7.2f} ms run, {(et-st)*1000.0:7.2f} ms python, {(cl-et)*1000.0:7.2f} ms {device_str}, {loss_cpu:7.2f} loss, {opt_non_bias.lr.numpy()[0]:.6f} LR, {GlobalCounters.mem_used/1e9:.2f} GB used, {GlobalCounters.global_ops*1e-9/(cl-st):9.2f} GFLOPS, {GlobalCounters.global_ops*1e-9:9.2f} GOPS")
      st = cl
      i += 1

+  if (assert_time:=getenv("ASSERT_MIN_STEP_TIME")):
+    min_time = min(step_times)
+    assert min_time < assert_time, f"Speed regression, expected min step time of < {assert_time} ms but took: {min_time} ms"
+
  # verify eval acc
  if target := getenv("TARGET_EVAL_ACC_PCT", 0.0):
    if eval_acc_pct >= target:
@@ -478,7 +478,7 @@ After you are done speaking, output [EOS]. You are not Chad.
      with Profiling(enabled=args.profile):
        with Timing("total ", enabled=args.timing, on_exit=lambda x: f", {1e9/x:.2f} tok/s, {GlobalCounters.global_mem/x:.2f} GB/s, param {param_bytes/x:.2f} GB/s"):
          with WallTimeEvent(BenchEvent.STEP):
-            with Timing("enqueue in ", on_exit=(lambda et: (f", {(GlobalCounters.time_sum_s-st)*1e3:.2f} ms on GPU" if DEBUG>=2 else "")+
+            with Timing("enqueue in ", on_exit=(lambda et: (f", {(GlobalCounters.time_sum_s-st)*1e3:.2f} ms on {Device.DEFAULT}" if DEBUG>=2 else "")+
                        f", {GlobalCounters.global_ops*1e-9:.2f} GOPS, {GlobalCounters.global_mem*1e-9:.2f} GB"+
                        (f", {GlobalCounters.global_mem*1e-9/(GlobalCounters.time_sum_s-st):.2f} GB/s, param {param_bytes*1e-9/(GlobalCounters.time_sum_s-st):.2f} GB/s" if DEBUG>=2 else "")) if DEBUG else None, enabled=args.timing):
              tok_tensor = llama.model(next_tok, start_pos, args.temperature)
@@ -441,7 +441,7 @@ if __name__ == "__main__":
      with Profiling(enabled=args.profile):
        with Timing("total ", on_exit=lambda x: f", {1e9/x:.2f} tok/s, {GlobalCounters.global_mem/x:.2f} GB/s, param {param_bytes/x:.2f} GB/s"):
          with WallTimeEvent(BenchEvent.STEP):
-            with Timing("enqueue in ", on_exit=(lambda et: (f", {(GlobalCounters.time_sum_s-st)*1e3:.2f} ms on GPU" if DEBUG>=2 else "")+
+            with Timing("enqueue in ", on_exit=(lambda et: (f", {(GlobalCounters.time_sum_s-st)*1e3:.2f} ms on {Device.DEFAULT}" if DEBUG>=2 else "")+
                        f", {GlobalCounters.global_ops*1e-9:.2f} GOPS, {GlobalCounters.global_mem*1e-9:.2f} GB"+
                        (f", {GlobalCounters.global_mem*1e-9/(GlobalCounters.time_sum_s-st):.2f} GB/s, param {param_bytes*1e-9/(GlobalCounters.time_sum_s-st):.2f} GB/s" if DEBUG>=2 else "")) if DEBUG else None):
              tok = model(Tensor([[last_tok]], device=device), start_pos, TEMPERATURE, TOP_K, TOP_P, ALPHA_F, ALPHA_P)
@@ -479,7 +479,7 @@ if __name__ == "__main__":
        st = GlobalCounters.time_sum_s
        with Profiling(enabled=args.profile):
          with Timing("total ", enabled=args.timing, on_exit=lambda x: f", {1e9/x:.2f} tok/s, {GlobalCounters.global_mem/x:.2f} GB/s, param {param_bytes/x:.2f} GB/s"):
-            with Timing("enqueue in ", on_exit=(lambda et: (f", {(GlobalCounters.time_sum_s-st)*1e3:.2f} ms on GPU" if DEBUG>=2 else "")+
+            with Timing("enqueue in ", on_exit=(lambda et: (f", {(GlobalCounters.time_sum_s-st)*1e3:.2f} ms on {Device.DEFAULT}" if DEBUG>=2 else "")+
                        f", {GlobalCounters.global_ops*1e-9:.2f} GOPS, {GlobalCounters.global_mem*1e-9:.2f} GB"+
                        (f", {GlobalCounters.global_mem*1e-9/(GlobalCounters.time_sum_s-st):.2f} GB/s, param {param_bytes*1e-9/(GlobalCounters.time_sum_s-st):.2f} GB/s" if DEBUG>=2 else "")) if DEBUG else None, enabled=args.timing):

@@ -279,9 +279,15 @@ def generate(model, tokenizer, prompt: str, n_tokens_to_gen: int = 10, temp: boo
  # Loading in the prompt tokens
  logits = model.forward(Tensor([tks]))[:, -1, :]
  for _ in tqdm(range(n_tokens_to_gen), desc="Speed Gen"):
-    # TODO: topk
    if sample:
-      tok_Tens = (logits/temp).softmax().multinomial()
+      scaled_logits = logits / temp
+      if top_k is not None:
+        topk_values, topk_indices = scaled_logits.topk(top_k)
+        filtered_logits = Tensor.full_like(scaled_logits, -float("inf"))
+        filtered_logits = filtered_logits.scatter(dim=-1, index=topk_indices, src=topk_values)
+        tok_Tens = filtered_logits.softmax().multinomial()
+      else:
+        tok_Tens = scaled_logits.softmax().multinomial()
    else:
      tok_Tens = logits.argmax(axis=-1).unsqueeze(0)
    tok = tok_Tens.item()
@@ -298,6 +304,7 @@ if __name__ == "__main__":
  parser.add_argument("--size", type=str, default="370m",
                      help=f"Size of model to use [{', '.join([k for k in MODELS.keys()])}]")
  parser.add_argument("--n_tokens", type=int, default=10, help="Number of tokens to generate")
+  parser.add_argument("--top_k", type=int, help="Limit sampling to the top k most likely tokens")
  parser.add_argument("--sample", dest="sample", action="store_true", help="Sample flag")
  parser.add_argument("--temp", type=float, default=1.0, help="Sampling temp has to be <=1.0")
  args = parser.parse_args()
@@ -308,8 +315,9 @@ if __name__ == "__main__":
  num_toks = args.n_tokens
  sample = args.sample
  temp = args.temp
+  top_k = args.top_k
  s = time.time()
-  tinyoutput = generate(model, tokenizer, prompt, n_tokens_to_gen=num_toks, sample=sample, temp=temp)
+  tinyoutput = generate(model, tokenizer, prompt, n_tokens_to_gen=num_toks, sample=sample, temp=temp, top_k=top_k)
  print(tinyoutput)
  print('TIME: ', time.time() - s)
  TORCHOUTPUT = "Why is gravity \nso important?\nBecause it's the only"
@@ -511,6 +511,33 @@ def batch_load_retinanet(dataset, val:bool, base_dir:Path, batch_size:int=32, sh
      # happens with BENCHMARK set
      pass

+# stable diffusion callbacks to match mlperf ref; declared here because they're pickled
+def filter_dataset(sample:dict): return {k:v for k,v in sample.items() if k in {'npy', 'txt'}}
+def collate(batch:list[dict]):
+  ret = {"npy": [], "txt": [], "__key__": []}
+  for sample in batch:
+    for k,v in sample.items():
+      ret[k].append(v)
+  return ret
+def collate_fn(batch): return batch
+
+# Reference (code): https://github.com/mlcommons/training/blob/2f4a93fb4888180755a8ef55f4b977ef8f60a89e/stable_diffusion/ldm/data/webdatasets.py, Line 55
+# Reference (params): https://github.com/mlcommons/training/blob/ab4ae1ca718d7fe62c369710a316dff18768d04b/stable_diffusion/configs/train_01x08x08.yaml, Line 107
+def batch_load_train_stable_diffusion(urls:str, BS:int):
+  import webdataset
+  dataset = webdataset.WebDataset(urls=urls, resampled=True, cache_size=-1, cache_dir=None)
+  dataset = dataset.shuffle(size=1000)
+  dataset = dataset.decode()
+  dataset = dataset.map(filter_dataset)
+  dataset = dataset.batched(BS, partial=False, collation_fn=collate)
+  dataset = webdataset.WebLoader(dataset, batch_size=None, shuffle=False, num_workers=1, persistent_workers=True, collate_fn=collate_fn)
+
+  for x in dataset:
+    assert isinstance(x, dict) and all(isinstance(k, str) for k in x.keys()) and all(isinstance(v, list) for v in x.values())
+    assert all(isinstance(moment_mean_logvar, np.ndarray) and moment_mean_logvar.shape==(1,8,64,64) for moment_mean_logvar in x["npy"])
+    assert all(isinstance(caption, str) for caption in x["txt"])
+    yield x
+
 # llama3

 class BinIdxDataset:
@@ -758,6 +785,27 @@ def batch_load_llama3(bs:int, samples:int, seqlen:int, base_dir:Path, seed:int=0
      batch.append(tokens)
    yield Tensor.stack(batch, dim=0)

+def batch_load_llama3_small(bs:int, samples:int, seqlen:int, base_dir:Path, seed:int=0, val:bool=True):
+  if val:
+    dataset = BlendedGPTDataset([
+      base_dir / "c4-validation-91205-samples.en_text_document",
+    ], [
+      1.0
+    ], samples, seqlen, seed, False)
+  else:
+    dataset = BlendedGPTDataset([
+      base_dir / "c4-train.en_6_text_document",
+    ], [
+      1.0
+    ], samples, seqlen, seed, True)
+
+  for b in range(math.ceil(samples / bs)):
+    batch = []
+    for i in range(bs):
+      tokens = dataset.get(b * bs + i)
+      batch.append(tokens)
+    yield Tensor.stack(batch, dim=0)
+
 if __name__ == "__main__":
  def load_unet3d(val):
    assert not val, "validation set is not supported due to different sizes on inputs"
@@ -2,7 +2,9 @@ import math
 from typing import Union

 from tinygrad import Tensor, nn, dtypes
-from tinygrad.helpers import prod, argfix
+from tinygrad.helpers import prod, argfix, Context
+from tinygrad.nn.state import get_parameters
+from extra.models.unet import UNetModel

 # rejection sampling truncated randn
 def rand_truncn(*shape, dtype=None, truncstds=2, **kwargs) -> Tensor:
@@ -17,6 +19,10 @@ def he_normal(*shape, a: float = 0.00, **kwargs) -> Tensor:
  std = math.sqrt(2.0 / (1 + a ** 2)) / math.sqrt(prod(argfix(*shape)[1:])) / 0.87962566103423978
  return std * rand_truncn(*shape, **kwargs)

+# Stable Diffusion v2 training uses default torch gelu, which doesn't use tanh approximation
+def gelu_erf(x:Tensor) -> Tensor:
+  return 0.5 * x * (1.0 + (x / 1.4142135623730951).erf())
+
 class Conv2dHeNormal(nn.Conv2d):
  def __init__(self, in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True):
    super().__init__(in_channels, out_channels, kernel_size, stride=stride, padding=padding, dilation=dilation, groups=groups, bias=bias)
@@ -127,3 +133,59 @@ class Conv2dRetinaNet(nn.Conv2d):
  def __call__(self, x:Tensor) -> Tensor:
    return x.conv2d(self.weight.cast(dtypes.default_float), self.bias.cast(dtypes.default_float) if self.bias is not None else None,
                    groups=self.groups, stride=self.stride, dilation=self.dilation, padding=self.padding)
+
+# copy torch AMP: isolate mixed precision to just the below autocast ops, instead of using dtypes.default_float which affects all new Tensors
+class AutocastLinear(nn.Linear):
+  cast_dtype=dtypes.bfloat16 # enable monkeypatching of the mixed precision dtype
+  def __call__(self, x:Tensor) -> Tensor:
+    dtype = type(self).cast_dtype
+    return x.cast(dtype).linear(self.weight.cast(dtype).transpose(), self.bias.cast(dtype) if self.bias is not None else None)
+
+class AutocastConv2d(nn.Conv2d):
+  cast_dtype=dtypes.bfloat16
+  def __call__(self, x:Tensor) -> Tensor:
+    dtype = type(self).cast_dtype
+    return x.cast(dtype).conv2d(self.weight.cast(dtype), self.bias.cast(dtype), self.groups, self.stride, self.dilation, self.padding)
+
+# copy torch AMP: upcast to float32 before GroupNorm and LayerNorm
+class AutocastGroupNorm(nn.GroupNorm):
+  def __call__(self, x:Tensor) -> Tensor:
+    return super().__call__(x.cast(dtypes.float32))
+
+class AutocastLayerNorm(nn.LayerNorm):
+  def __call__(self, x:Tensor) -> Tensor:
+    return super().__call__(x.cast(dtypes.float32))
+
+def zero_module(module):
+  for p in get_parameters(module): p.assign(Tensor.zeros_like(p).contiguous())
+
+# Stable Diffusion mlperf reference doesn't call scaled_dot_product_attention
+# copy torch AMP: upcast to float32 before softmax on CUDA
+def attn_f32_softmax(q:Tensor, k:Tensor, v:Tensor) -> Tensor:
+  return (q.matmul(k.transpose(-2,-1), dtype=dtypes.float32) / math.sqrt(q.shape[-1])).softmax(-1).cast(q.dtype) @ v
+
+def init_stable_diffusion(version:str, pretrained:str, devices:list[str]):
+  from examples.stable_diffusion import StableDiffusion
+  from tinygrad.nn.state import safe_load, safe_save, load_state_dict, get_state_dict
+  from tempfile import TemporaryDirectory
+  model = StableDiffusion(version=version, pretrained=pretrained)
+  unet:UNetModel = model.model.diffusion_model
+
+  # this prevents extra consumption of memory, enabling much larger BS
+  Tensor.realize(*get_parameters(unet))
+  with TemporaryDirectory(prefix="unet_init") as tmp:
+    safe_save(get_state_dict(unet), init_fn:=f"{tmp}/init_model.safetensors")
+    load_state_dict(unet, safe_load(init_fn))
+
+  sqrt_alphas_cumprod = model.alphas_cumprod.sqrt().realize()
+  sqrt_one_minus_alphas_cumprod = (1 - model.alphas_cumprod).sqrt().realize()
+
+  if len(devices) > 1:
+    to_move = [sqrt_alphas_cumprod, sqrt_one_minus_alphas_cumprod]
+    if version == "v2-mlperf-train": to_move += get_parameters(unet) + get_parameters(model.cond_stage_model)
+    for p in to_move:
+      p.to_(devices)
+    with Context(BEAM=0):
+      Tensor.realize(*to_move)
+
+  return model, unet, sqrt_alphas_cumprod, sqrt_one_minus_alphas_cumprod
@@ -1,8 +1,9 @@
 import math
-from tinygrad import dtypes
+from tinygrad import dtypes, Tensor
 from tinygrad.nn.optim import Optimizer

 from extra.lr_scheduler import LR_Scheduler
+from typing import Callable

 # https://github.com/mlcommons/training/blob/e237206991d10449d9675d95606459a3cb6c21ad/image_classification/tensorflow2/lars_util.py
 class PolynomialDecayWithWarmup(LR_Scheduler):
@@ -36,4 +37,24 @@ class CosineAnnealingLRWithWarmup(LR_Scheduler):
  def get_lr(self):
    warmup_lr = ((self.epoch_counter+1) / self.warmup_steps) * self.base_lr
    decay_lr = self.end_lr + 0.5 * (self.base_lr-self.end_lr) * (1 + (((self.epoch_counter+1-self.warmup_steps)/self.decay_steps) * math.pi).cos())
-    return (self.epoch_counter < self.warmup_steps).where(warmup_lr, decay_lr).cast(self.optimizer.lr.dtype)
+    return (self.epoch_counter < self.warmup_steps).where(warmup_lr, decay_lr).cast(self.optimizer.lr.dtype)
+
+# Reference: https://github.com/mlcommons/training/blob/64b14a9abc74e08779a175abca7d291f8c957632/stable_diffusion/ldm/lr_scheduler.py, Lines 36-97
+class LambdaLinearScheduler:
+  def __init__(self, warm_up_steps:int, f_min:float, f_max:float, f_start:float, cycle_lengths:int):
+    self.lr_warm_up_steps, self.f_min, self.f_max, self.f_start, self.cycle_lengths = warm_up_steps, f_min, f_max, f_start, cycle_lengths
+
+  def schedule(self, n:Tensor) -> Tensor:
+    warm_up = (n < self.lr_warm_up_steps)
+    f_warm_up = (self.f_max - self.f_start) / self.lr_warm_up_steps * n + self.f_start
+    return warm_up.where(f_warm_up, self.f_min + (self.f_max - self.f_min) * (self.cycle_lengths - n) / (self.cycle_lengths))
+
+# based on torch.optim.lr_scheduler.LambdaLR
+class LambdaLR(LR_Scheduler):
+  def __init__(self, optimizer:Optimizer, base_lr:Tensor, lr_lambda:Callable):
+    super().__init__(optimizer)
+    self.base_lr, self.lr_lambda = base_lr, lr_lambda
+    self.step()
+
+  def get_lr(self):
+    return self.base_lr * self.lr_lambda(self.epoch_counter - 1)
@@ -1,10 +1,10 @@
-import time, math
+import time, math, os
 start = time.perf_counter()
 from pathlib import Path
 import numpy as np
 from tinygrad import Tensor, Device, dtypes, GlobalCounters, TinyJit
 from tinygrad.nn.state import get_parameters, load_state_dict, safe_load
-from tinygrad.helpers import getenv
+from tinygrad.helpers import getenv, Context, prod
 from extra.bench_log import BenchEvent, WallTimeEvent
 def tlog(x): print(f"{x:25s}  @ {time.perf_counter()-start:5.2f}s")

@@ -243,31 +243,299 @@ def eval_mrcnn():

 def eval_llama3():
  from extra.models.llama import Transformer
-  from examples.llama3 import MODEL_PARAMS
+  from examples.llama3 import MODEL_PARAMS, load, convert_from_huggingface
  from tinygrad.helpers import tqdm

-  bs = 4
-  sequence_length = 512
+  BASEDIR = Path(getenv("BASEDIR", "/raid/datasets/c4/"))
+  BS = getenv("BS", 4)
+  SMALL = getenv("SMALL", 0)
+  SEQLEN = getenv("SEQLEN", 8192)
+  MODEL_PATH = Path(getenv("MODEL_PATH", "/raid/weights/llama31_8b/"))

-  model = Transformer(**(MODEL_PARAMS[getenv("LLAMA3_SIZE", "8B")]["args"]|{"vocab_size": 32000}), max_context=sequence_length, jit=False, disable_kv_cache=True)
+  params = MODEL_PARAMS[getenv("LLAMA3_SIZE", "8B")]["args"]
+  params = params | {"vocab_size": 32000} if not SMALL else params
+  if (llama_layers:=getenv("LLAMA_LAYERS")) != 0: params['n_layers'] = llama_layers
+  model = Transformer(**params, max_context=SEQLEN, jit=False, disable_kv_cache=True)
+
+  # load weights
+  weights = load(str(MODEL_PATH / "model.safetensors.index.json"))
+  if "model.embed_tokens.weight" in weights:
+    print("converting from huggingface format")
+    weights = convert_from_huggingface(weights, params["n_layers"], params["n_heads"], params["n_kv_heads"])
+
+  load_state_dict(model, weights, strict=False, consume=True)

  @TinyJit
  def eval_step(model, tokens):
    logits:Tensor = model(tokens[:, :-1], start_pos=0, temperature=math.nan)
    loss = logits.sparse_categorical_crossentropy(tokens[:, 1:])
-    return loss.flatten()
+    return loss.flatten().float()

-  from examples.mlperf.dataloader import batch_load_llama3
-  iter = batch_load_llama3(bs, 5760, sequence_length, Path(getenv("BASEDIR", "/raid/datasets/c4/")), True)
+  if SMALL:
+    from examples.mlperf.dataloader import batch_load_llama3_small
+    iter = batch_load_llama3_small(BS, 5760, SEQLEN, BASEDIR, val=True)
+  else:
+    from examples.mlperf.dataloader import batch_load_llama3
+    iter = batch_load_llama3(BS, 5760, SEQLEN, BASEDIR, val=True)

  losses = []
-  for tokens in tqdm(iter, total=5760//bs):
+  for tokens in tqdm(iter, total=5760//BS):
    GlobalCounters.reset()
    losses += eval_step(model, tokens).tolist()
    tqdm.write(f"loss: {np.mean(losses)}")

-  log_perplexity = Tensor(losses).mean()
-  print(f"Log Perplexity: {log_perplexity.item()}")
+  log_perplexity = np.mean(losses)
+  print(f"Log Perplexity: {log_perplexity}")
+
+# NOTE: BEAM hangs on 8xmi300x with DECODE_BS=384 in final realize below; function is declared here for external testing
+@TinyJit
+def vae_decode(x:Tensor, vae, disable_beam=False) -> Tensor:
+  from examples.stable_diffusion import AutoencoderKL
+  assert isinstance(vae, AutoencoderKL)
+  x = vae.post_quant_conv(1./0.18215 * x)
+
+  x = vae.decoder.conv_in(x)
+  x = vae.decoder.mid(x)
+  for i, l in enumerate(vae.decoder.up[::-1]):
+    print("decode", x.shape)
+    for b in l['block']: x = b(x)
+    if 'upsample' in l:
+      bs,c,py,px = x.shape
+      x = x.reshape(bs, c, py, 1, px, 1).expand(bs, c, py, 2, px, 2).reshape(bs, c, py*2, px*2)
+      x = l['upsample']['conv'](x)
+    if i == len(vae.decoder.up) - 1 and disable_beam:
+      with Context(BEAM=0): x.realize()
+    else: x.realize()
+  x = vae.decoder.conv_out(vae.decoder.norm_out(x).swish())
+
+  x = ((x + 1.0) / 2.0).clip(0.0, 1.0)
+  return x
+
+def eval_stable_diffusion():
+  import csv, PIL, sys
+  from tqdm import tqdm
+  from examples.mlperf.initializers import init_stable_diffusion, gelu_erf
+  from examples.stable_diffusion import AutoencoderKL
+  from extra.models.unet import UNetModel
+  from tinygrad.nn.state import load_state_dict, torch_load
+  from tinygrad.helpers import BEAM
+  from extra.models import clip
+  from extra.models.clip import FrozenOpenClipEmbedder
+  from extra.models.clip import OpenClipEncoder
+  from extra.models.inception import FidInceptionV3
+
+  config = {}
+  GPUS               = config["GPUS"]                   = [f"{Device.DEFAULT}:{i}" for i in range(getenv("GPUS", 1))]
+  for x in GPUS: Device[x]
+  print(f"running eval on {GPUS}")
+  seed               = config["seed"]                   = getenv("SEED", 12345)
+  CKPTDIR            = config["CKPTDIR"]                = Path(getenv("CKPTDIR", "./checkpoints"))
+  DATADIR            = config["DATADIR"]                = Path(getenv("DATADIR", "./datasets"))
+  CONTEXT_BS         = config["CONTEXT_BS"]             = getenv("CONTEXT_BS", 1 * len(GPUS))
+  DENOISE_BS         = config["DENOISE_BS"]             = getenv("DENOISE_BS", 1 * len(GPUS))
+  DECODE_BS          = config["DECODE_BS"]              = getenv("DECODE_BS", 1 * len(GPUS))
+  INCEPTION_BS       = config["INCEPTION_BS"]           = getenv("INCEPTION_BS", 1 * len(GPUS))
+  CLIP_BS            = config["CLIP_BS"]                = getenv("CLIP_BS", 1 * len(GPUS))
+  EVAL_CKPT_DIR      = config["EVAL_CKPT_DIR"]          = getenv("EVAL_CKPT_DIR", "")
+  STOP_IF_CONVERGED  = config["STOP_IF_CONVERGED"]      = getenv("STOP_IF_CONVERGED", 0)
+
+  if (WANDB := getenv("WANDB", "")):
+    import wandb
+    wandb.init(config=config, project="MLPerf-Stable-Diffusion")
+
+  assert EVAL_CKPT_DIR != "", "provide a directory with checkpoints to be evaluated"
+  print(f"running eval on checkpoints in {EVAL_CKPT_DIR}\nSEED={seed}")
+  eval_queue:list[tuple[int, Path]] = []
+  for p in Path(EVAL_CKPT_DIR).iterdir():
+    if p.name.endswith(".safetensors"):
+      ckpt_iteration = p.name.split(".safetensors")[0]
+      assert ckpt_iteration.isdigit(), f"invalid checkpoint name: {p.name}, expected <digits>.safetensors"
+      eval_queue.append((int(ckpt_iteration), p))
+  assert len(eval_queue), f'no files ending with ".safetensors" were found in {EVAL_CKPT_DIR}'
+  print(sorted(eval_queue, reverse=True))
+
+  Tensor.manual_seed(seed)  # seed for weight initialization
+  model, unet, sqrt_alphas_cumprod, sqrt_one_minus_alphas_cumprod = init_stable_diffusion("v2-mlperf-eval", CKPTDIR / "sd" / "512-base-ema.ckpt", GPUS)
+
+  # load prompts for generating images for validation; 2 MB of data total
+  with open(DATADIR / "coco2014" / "val2014_30k.tsv") as f:
+    reader = csv.DictReader(f, delimiter="\t")
+    eval_inputs:list[dict] = [{"image_id": int(row["image_id"]), "id": int(row["id"]), "caption": row["caption"]} for row in reader]
+  assert len(eval_inputs) == 30_000
+  # NOTE: the clip weights are the same between model.cond_stage_model and clip_encoder
+  eval_timesteps = list(reversed(range(1, 1000, 20)))
+
+  original_device, Device.DEFAULT = Device.DEFAULT, "CPU"
+  # The choice of alphas_prev[0] = alphas_cumprod[0] seems arbitrary, but it's how the mlperf ref does it:
+  #   alphas_prev = np.asarray([alphacums[0]] + alphacums[ddim_timesteps[:-1]].tolist())
+  eval_alphas_prev = model.alphas_cumprod[0:1].cat(model.alphas_cumprod[list(range(1, 1000, 20))[:-1]]).to(GPUS).realize()
+  inception = FidInceptionV3().load_from_pretrained(CKPTDIR / "inception" / "pt_inception-2015-12-05-6726825d.pth")
+  vision_cfg = {'width': 1280, 'layers': 32, 'd_head': 80, 'image_size': 224, 'patch_size': 14}
+  text_cfg = {'width': 1024, 'n_heads': 16, 'layers': 24, 'vocab_size': 49408, 'ctx_length': 77}
+  clip.gelu = gelu_erf
+  clip_encoder = OpenClipEncoder(1024, text_cfg, vision_cfg)
+  loaded = torch_load(CKPTDIR / "clip" / "open_clip_pytorch_model.bin")
+  loaded.update({"attn_mask": clip_encoder.attn_mask, "mean": clip_encoder.mean, "std": clip_encoder.std})
+  load_state_dict(clip_encoder, loaded)
+  Device.DEFAULT=original_device
+
+  @TinyJit
+  def denoise_step(x:Tensor, x_x:Tensor, t_t:Tensor, uc_c:Tensor, sqrt_alphas_cumprod_t:Tensor, sqrt_one_minus_alphas_cumprod_t:Tensor,
+                    alpha_prev:Tensor, unet:UNetModel, GPUS) -> Tensor:
+    out_uncond, out = unet(x_x, t_t, uc_c).to("CPU").reshape(-1, 2, 4, 64, 64).chunk(2, dim=1)
+    out_uncond = out_uncond.squeeze(1).shard(GPUS,axis=0)
+    out = out.squeeze(1).shard(GPUS,axis=0)
+    v_t = out_uncond + 8.0 * (out - out_uncond)
+    e_t = sqrt_alphas_cumprod_t * v_t + sqrt_one_minus_alphas_cumprod_t * x
+    pred_x0 = sqrt_alphas_cumprod_t * x - sqrt_one_minus_alphas_cumprod_t * v_t
+    dir_xt = (1. - alpha_prev).sqrt() * e_t
+    x_prev = alpha_prev.sqrt() * pred_x0 + dir_xt
+    return x_prev.realize()
+
+  def shard_tensor(t:Tensor) -> Tensor: return t.shard(GPUS, axis=0) if len(GPUS) > 1 else t.to(GPUS[0])
+  def get_batch(whole:Tensor, i:int, bs:int) -> tuple[Tensor, int]:
+    batch = whole[i: i + bs].to("CPU")
+    if (unpadded_bs:=batch.shape[0]) < bs:
+      batch = batch.cat(batch[-1:].expand(bs - unpadded_bs, *batch[-1].shape))
+    return batch, unpadded_bs 
+
+  @Tensor.train(mode=False)
+  def eval_unet(eval_inputs:list[dict], unet:UNetModel, cond_stage:FrozenOpenClipEmbedder, first_stage:AutoencoderKL,
+                inception:FidInceptionV3, clip:OpenClipEncoder) -> tuple[float, float]:
+    # Eval is divided into 5 jits, one per model
+    # It doesn't make sense to merge these jits, e.g. unet repeats 50 times in isolation; images fork to separate inception/clip
+    # We're generating and scoring 30,000 images per eval, and all the data can flow through one jit at a time
+    # To maximize throughput for each jit, we have only one model/jit on the GPU at a time, and pool outputs from each jit off-GPU
+    for model in (unet, first_stage, inception, clip):
+      Tensor.realize(*[p.to_("CPU") for p in get_parameters(model)])
+
+    uc_written = False
+    models = (cond_stage, unet, first_stage, inception, clip)
+    jits = (jit_context:=TinyJit(cond_stage.embed_tokens), denoise_step, vae_decode, jit_inception:=TinyJit(inception),
+            jit_clip:=TinyJit(clip.get_clip_score))
+    all_bs = (CONTEXT_BS, DENOISE_BS, DECODE_BS, INCEPTION_BS, CLIP_BS)
+    if (EVAL_SAMPLES:=getenv("EVAL_SAMPLES", 0)) and EVAL_SAMPLES > 0:
+      eval_inputs = eval_inputs[0:EVAL_SAMPLES]
+    output_shapes = [(ns:=len(eval_inputs),77), (ns,77,1024), (ns,4,64,64), (ns,3,512,512), (ns,2048), (ns,)]
+    # Writing progress to disk lets us resume eval if we crash
+    stages = ["tokens", "embeds", "latents", "imgs", "inception", "clip"]
+    disk_tensor_names, disk_tensor_shapes = stages + ["end", "uc"], output_shapes + [(6,), (1,77,1024)]
+    if not all(os.path.exists(f"{EVAL_CKPT_DIR}/{name}.bytes") for name in disk_tensor_names):
+      for name, shape in zip(disk_tensor_names, disk_tensor_shapes):
+        file = Path(f"{EVAL_CKPT_DIR}/{name}.bytes")
+        file.unlink(missing_ok=True)
+        with file.open("wb") as f: f.truncate(prod(shape) * 4)
+    progress = {name: Tensor.empty(*shape, device=f"disk:{EVAL_CKPT_DIR}/{name}.bytes", dtype=dtypes.int if name in {"tokens", "end"} else dtypes.float)
+                for name, shape in zip(disk_tensor_names, disk_tensor_shapes)}
+
+    def embed_tokens(tokens:Tensor) -> Tensor:
+      nonlocal uc_written
+      if not uc_written:
+        with Context(BEAM=0): progress["uc"].assign(cond_stage.embed_tokens(cond_stage.tokenize("").to(GPUS)).to("CPU").realize()).realize()
+        uc_written = True
+      return jit_context(shard_tensor(tokens))
+
+    def generate_latents(embeds:Tensor) -> Tensor:
+      uc_c = Tensor.stack(progress["uc"].to("CPU").expand(bs, 77, 1024), embeds, dim=1).reshape(-1, 77, 1024)
+      uc_c = shard_tensor(uc_c)
+      x = shard_tensor(Tensor.randn(bs,4,64,64))
+      for step_idx, timestep in enumerate(tqdm(eval_timesteps)):
+        reversed_idx = Tensor([50 - step_idx - 1], device=GPUS)
+        alpha_prev = eval_alphas_prev[reversed_idx]
+        ts = Tensor.full(bs, fill_value=timestep, dtype=dtypes.int, device="CPU")
+        ts_ts = shard_tensor(ts.cat(ts))
+        ts = shard_tensor(ts)
+        sqrt_alphas_cumprod_t = sqrt_alphas_cumprod[ts].reshape(bs, 1, 1, 1)
+        sqrt_one_minus_alphas_cumprod_t = sqrt_one_minus_alphas_cumprod[ts].reshape(bs, 1, 1, 1)
+        x_x = shard_tensor(Tensor.stack(x.to("CPU"), x.to("CPU"), dim=1).reshape(-1, 4, 64, 64))
+        x.assign(denoise_step(x, x_x, ts_ts, uc_c, sqrt_alphas_cumprod_t, sqrt_one_minus_alphas_cumprod_t, alpha_prev, unet, GPUS)).realize()
+      return x
+
+    def decode_latents(latents:Tensor) -> Tensor: return vae_decode(shard_tensor(latents), first_stage, disable_beam=True)
+    def generate_inception(imgs:Tensor) -> Tensor: return jit_inception(shard_tensor(imgs))[:,:,0,0]
+
+    def calc_clip_scores(batch:Tensor, batch_tokens:Tensor) -> Tensor:
+      # Tensor.interpolate does not yet support bicubic, so we use PIL
+      batch = (batch.to(GPUS[0]).permute(0,2,3,1) * 255).clip(0, 255).cast(dtypes.uint8).numpy()
+      batch = [np.array(PIL.Image.fromarray(batch[i]).resize((224,224), PIL.Image.BICUBIC)) for i in range(bs)]
+      batch = shard_tensor(Tensor(np.stack(batch, axis=0).transpose(0,3,1,2), device="CPU").realize())
+      batch = batch.cast(dtypes.float) / 255
+      batch = (batch - model.mean) / model.std
+      batch = jit_clip(shard_tensor(batch_tokens), batch)
+      return batch
+
+    callbacks = (embed_tokens, generate_latents, decode_latents, generate_inception, calc_clip_scores)
+
+    # save every forward pass output to disk; NOTE: this needs ~100 GB disk space because 30k images are large
+    def stage_progress(stage_idx:int) -> int: return progress["end"].to("CPU")[stage_idx].item()
+    if stage_progress(0) < len(eval_inputs):
+      tokens = []
+      for i in tqdm(range(0, len(eval_inputs), CONTEXT_BS)):
+        subset = [cond_stage.tokenize(row["caption"], device="CPU") for row in eval_inputs[i: i+CONTEXT_BS]]
+        tokens.append(Tensor.cat(*subset, dim=0).realize())
+      progress["tokens"].assign(Tensor.cat(*tokens, dim=0).realize()).realize()
+      progress["end"][0:1].assign(Tensor([len(eval_inputs)], dtype=dtypes.int)).realize()
+    prev_stage = "tokens"
+    tokens = progress["tokens"]
+
+    # wrapper code for every model
+    for stage_idx, model, jit, bs, callback in zip(range(1,6), models, jits, all_bs, callbacks):
+      stage = stages[stage_idx]
+      if stage_progress(stage_idx) >= len(eval_inputs):
+        prev_stage = stage
+        continue # use cache
+      t0 = time.perf_counter()
+      print(f"starting eval with model: {model}")
+      if stage_idx == 1: inputs = tokens
+      elif stage_idx == 5: inputs = progress["imgs"]
+      else: inputs = progress[prev_stage]
+
+      Tensor.realize(*[p.to_(GPUS) for p in get_parameters(model)])
+      for batch_idx in tqdm(range(stage_progress(stage_idx), inputs.shape[0], bs)):
+        t1 = time.perf_counter()
+        batch, unpadded_bs = get_batch(inputs, batch_idx, bs)
+        if isinstance(model, OpenClipEncoder): batch = callback(batch, get_batch(tokens, batch_idx, bs)[0].realize())
+        else: batch = callback(batch)
+        # to(GPUS[0]) is necessary for this to work, without that the result is still on GPUS, probably due to a bug
+        batch = batch.to(GPUS[0]).to("CPU")[0:unpadded_bs].realize()
+        progress[stage][batch_idx: batch_idx + bs].assign(batch).realize()
+        # keep track of what our last output was, so we can resume from there if we crash in this loop
+        progress["end"][stage_idx: stage_idx + 1].assign(Tensor([batch_idx + bs], dtype=dtypes.int)).realize()
+        print(f"model: {model}, batch_idx: {batch_idx}, elapsed: {(time.perf_counter() - t1):.2f}")
+      del batch
+        
+      jit.reset()
+      Tensor.realize(*[p.to_("CPU") for p in get_parameters(model)])
+      print(f"done with model: {model}, elapsed: {(time.perf_counter() - t0):.2f}")
+      prev_stage = stage
+
+    inception_stats_fn = str(DATADIR / "coco2014" / "val2014_30k_stats.npz")
+    fid_score = inception.compute_score(progress["inception"].to("CPU"), inception_stats_fn)
+    clip_score = progress["clip"].to(GPUS[0]).mean().item()
+    for name in disk_tensor_names:
+      Path(f"{EVAL_CKPT_DIR}/{name}.bytes").unlink(missing_ok=True)
+    
+    if EVAL_SAMPLES and BEAM:
+      print("BEAM COMPLETE", flush=True) # allows wrapper script to detect BEAM search completion and retry if it failed
+      sys.exit() # Don't eval additional models; we don't care about clip/fid scores when running BEAM on eval sample subset
+
+    return clip_score, fid_score
+
+  # evaluate checkpoints in reverse chronological order
+  for ckpt_iteration, p in sorted(eval_queue, reverse=True):
+    unet_ckpt = safe_load(p)
+    load_state_dict(unet, unet_ckpt)
+    clip_score, fid_score = eval_unet(eval_inputs, unet, model.cond_stage_model, model.first_stage_model, inception, clip_encoder)
+    converged = True if clip_score >= 0.15 and fid_score <= 90 else False
+    print(f"eval results for {EVAL_CKPT_DIR}/{p.name}: clip={clip_score}, fid={fid_score}, converged={converged}")
+    if WANDB:
+      wandb.log({"eval/ckpt_iteration": ckpt_iteration, "eval/clip_score": clip_score, "eval/fid_score": fid_score})
+    if converged and STOP_IF_CONVERGED:
+      print(f"Convergence detected, exiting early before evaluating other checkpoints due to STOP_IF_CONVERGED={STOP_IF_CONVERGED}")
+      sys.exit()
+
+  # for testing
+  return clip_score, fid_score, ckpt_iteration

 if __name__ == "__main__":
  # inference only
@@ -3,8 +3,8 @@ from pathlib import Path
 import multiprocessing

 from tinygrad import Device, GlobalCounters, Tensor, TinyJit, dtypes
-from tinygrad.helpers import getenv, BEAM, WINO, round_up, diskcache_clear, FUSE_CONV_BW, Profiling
-from tinygrad.nn.state import get_parameters, get_state_dict, safe_load, safe_save
+from tinygrad.helpers import getenv, BEAM, WINO, round_up, diskcache_clear, Profiling
+from tinygrad.nn.state import get_parameters, get_state_dict, load_state_dict, safe_load, safe_save
 from tinygrad.nn.optim import LAMB, LARS, SGD, OptimizerGroup, Adam, AdamW

 from extra.lr_scheduler import LRSchedulerGroup
@@ -252,6 +252,10 @@ def train_resnet():
        print(f"epoch global_ops: {steps_in_train_epoch * GlobalCounters.global_ops:_}, "
              f"epoch global_mem: {steps_in_train_epoch * GlobalCounters.global_mem:_}")
        # if we are doing beam search, run the first eval too
+        if (assert_time:=getenv("ASSERT_MIN_STEP_TIME")):
+          min_time = min(step_times)
+          assert min_time < assert_time, f"Speed regression, expected min step time of < {assert_time} ms but took: {min_time} ms"
+
        if (TRAIN_BEAM or EVAL_BEAM) and e == start_epoch: break
        return
    if MLLOGGER and RUNMLPERF:
@@ -344,6 +348,8 @@ def train_resnet():
        print(f"saving ckpt to {fn}")
        safe_save(get_training_state(model, optimizer_group, scheduler_group), fn)

+
+
 def train_retinanet():
  from contextlib import redirect_stdout
  from examples.mlperf.dataloader import batch_load_retinanet
@@ -701,7 +707,7 @@ def train_unet3d():
  ```BASEDIR=<folder_path> ./examples/mlperf/scripts/setup_kits19_dataset.sh```

  2) To start training the model, run the following:
-  ```time PYTHONPATH=. WANDB=1 TRAIN_BEAM=3 FUSE_CONV_BW=1 GPUS=6 BS=6 MODEL=unet3d python3 examples/mlperf/model_train.py```
+  ```time PYTHONPATH=. WANDB=1 TRAIN_BEAM=3 GPUS=6 BS=6 MODEL=unet3d python3 examples/mlperf/model_train.py```
  """
  from examples.mlperf.losses import dice_ce_loss
  from examples.mlperf.metrics import dice_score
@@ -743,7 +749,6 @@ def train_unet3d():
    "train_beam": TRAIN_BEAM,
    "eval_beam": EVAL_BEAM,
    "wino": WINO.value,
-    "fuse_conv_bw": FUSE_CONV_BW.value,
    "gpus": GPUS,
    "default_float": dtypes.default_float.name
  }
@@ -1290,15 +1295,20 @@ def train_llama3():
  from examples.mlperf.lr_schedulers import CosineAnnealingLRWithWarmup

  config = {}
+  BASEDIR            = config["BASEDIR"]                = Path(getenv("BASEDIR", "/raid/datasets/c4/"))
  BS                 = config["BS"]                     = getenv("BS", 16)
  grad_acc           = config["GRADIENT_ACC_STEPS"]     = getenv("GRADIENT_ACC_STEPS", 1)
  GBS                = config["GLOBAL_BATCH_SIZE"]      = BS * grad_acc
  SEED               = config["SEED"]                   = getenv("SEED", 5760)
  SEQLEN             = config["SEQLEN"]                 = getenv("SEQLEN", 8192)
  TRAIN_ON_VAL       = config["TRAIN_ON_VAL"]           = getenv("TRAIN_ON_VAL", 0)
+  SMALL              = config["SMALL"]                  = getenv("SMALL", 0)
  SAMPLES            = config["SAMPLES"]                = getenv("SAMPLES", 5_760 if TRAIN_ON_VAL else 1_200_000 * 1152)
+  EVAL_FREQ          = config["EVAL_FREQ"]              = getenv("EVAL_FREQ", 46080)
+  EVAL_BS            = config["EVAL_BS"]                = getenv("EVAL_BS", 16)
+  EVAL_TARGET        = config["EVAL_TARGET"]            = getenv("EVAL_TARGET", 5.6)

-  # LR=1e-4 TRAIN_ON_VAL=1 DEFAULT_FLOAT=bfloat16 FUSE_ARANGE=1 JITBEAM=2 OPTIM_DTYPE=bfloat16 LLAMA3_SIZE=1B WARMUP_STEPS=36 DECAY_STEPS=360 SEQLEN=512 PYTHONPATH=. AMD=1 AMD_LLVM=0 MODEL=llama3 python3 examples/mlperf/model_train.py
+  # LR=1e-4 TRAIN_ON_VAL=1 DEFAULT_FLOAT=bfloat16 JITBEAM=2 OPTIM_DTYPE=bfloat16 LLAMA3_SIZE=1B WARMUP_STEPS=36 DECAY_STEPS=360 SEQLEN=512 PYTHONPATH=. AMD=1 AMD_LLVM=0 MODEL=llama3 python3 examples/mlperf/model_train.py
  # trains to 7

  opt_adamw_beta_1 = 0.9
@@ -1308,16 +1318,21 @@ def train_llama3():

  opt_gradient_clip_norm = 1.0
  opt_learning_rate_warmup_steps = getenv("WARMUP_STEPS", math.ceil(8000 * 1152 / GBS))
-  opt_learning_rate_decay_steps = getenv("DECAY_STEPS", math.ceil(1_200_000 * 1152 / GBS) - opt_learning_rate_warmup_steps)
+  opt_learning_rate_decay_steps = getenv("MAX_STEPS", math.ceil(1_200_000 * 1152 / GBS)) - opt_learning_rate_warmup_steps
  opt_base_learning_rate = getenv("LR", 8e-5 * GBS / 1152)  # NOTE: cannot change for benchmark
-  opt_end_learning_rate = 8e-7
+  opt_end_learning_rate = getenv("END_LR", 8e-7)

  # TODO: confirm weights are in bf16
  # vocab_size from the mixtral tokenizer
-  params = MODEL_PARAMS[getenv("LLAMA3_SIZE", "8B")]["args"]|{"vocab_size": 32000}
+  params = MODEL_PARAMS[getenv("LLAMA3_SIZE", "8B")]["args"]
+  params = params | {"vocab_size": 32000} if not SMALL else params
  if (llama_layers:=getenv("LLAMA_LAYERS")) != 0: params['n_layers'] = llama_layers
  model = Transformer(**params, max_context=SEQLEN, jit=False, disable_kv_cache=True)

+  if getenv("FAKEDATA"):
+    for v in get_parameters(model):
+      v = v.assign(Tensor.empty(v.shape))
+
  if (DP := getenv("DP", 1)) > 1:
    device = tuple(f"{Device.DEFAULT}:{i}" for i in range(DP))
    for v in get_parameters(model):
@@ -1339,11 +1354,22 @@ def train_llama3():
      else:
        # attention_norm, ffn_norm, norm
        v.shard_(device, axis=None)
+      # prevents memory spike on device 0
+      v.realize()

  optim = AdamW(get_parameters(model), lr=0.0,
                b1=opt_adamw_beta_1, b2=opt_adamw_beta_2, eps=opt_adamw_epsilon, weight_decay=opt_adamw_weight_decay)
  scheduler = CosineAnnealingLRWithWarmup(optim, opt_base_learning_rate, opt_end_learning_rate, opt_learning_rate_warmup_steps, opt_learning_rate_decay_steps)

+  if resume_ckpt := getenv("RESUME_CKPT"):
+    fn = f"./ckpts/llama3_{resume_ckpt}.safe"
+    print(f"loading initial checkpoint from {fn}")
+    load_state_dict(model, safe_load(fn), realize=False)
+
+    fn = f"./ckpts/llama3_{resume_ckpt}_optim.safe"
+    print(f"loading optim checkpoint from {fn}")
+    load_state_dict(scheduler, safe_load(fn), realize=False)
+
  @TinyJit
  @Tensor.train()
  def train_step(model, tokens:Tensor, grad_acc:int):
@@ -1369,7 +1395,7 @@ def train_llama3():
        total_norm += p.grad.float().square().sum()
      total_norm = total_norm.sqrt().contiguous()
      for p in optim.params:
-        p.grad = p.grad * opt_gradient_clip_norm / (total_norm + 1e-6)
+        p.grad = p.grad * (opt_gradient_clip_norm / (total_norm + 1e-6)).clamp(max_=1.0)

    optim.step()
    scheduler.step()
@@ -1378,33 +1404,231 @@ def train_llama3():
    loss.realize(lr)
    return loss, lr

-  if getenv("FAKEDATA", 0):
-    def fake_data():
-      for _ in range(SAMPLES // GBS):
-        yield Tensor.randint(GBS, SEQLEN + 1, low=0, high=32000, dtype=dtypes.int32, device=Device.DEFAULT)
-    iter = fake_data()
-  else:
-    from examples.mlperf.dataloader import batch_load_llama3
-    iter = batch_load_llama3(GBS, SAMPLES, SEQLEN, Path(getenv("BASEDIR", "/raid/datasets/c4/")), seed=SEED, val=bool(TRAIN_ON_VAL))
+  @TinyJit
+  @Tensor.train(False)
+  def eval_step(model, tokens:Tensor):
+    if (DP := getenv("DP", 1)) > 1:
+      device = tuple(f"{Device.DEFAULT}:{i}" for i in range(DP))
+      tokens = tokens.shard(device, 0)
+    if (MP := getenv("MP", 1)) > 1:
+      device = tuple(f"{Device.DEFAULT}:{i}" for i in range(MP))
+      tokens = tokens.shard(device)
+    logits:Tensor = model(tokens[:, :-1], start_pos=0, temperature=math.nan)
+    loss = logits.sparse_categorical_crossentropy(tokens[:, 1:])
+    return loss.flatten().float()

-  i = 0
+  # ** data iters **
+  def fake_data(bs, samples):
+    for _ in range(samples // bs):
+      yield Tensor.randint(bs, SEQLEN + 1, low=0, high=params["vocab_size"], dtype=dtypes.int32, device=Device.DEFAULT)
+
+  def get_train_iter():
+    if getenv("FAKEDATA", 0):
+      return fake_data(GBS, SAMPLES)
+    else:
+      if SMALL:
+        from examples.mlperf.dataloader import batch_load_llama3_small
+        return batch_load_llama3_small(GBS, SAMPLES, SEQLEN, BASEDIR, seed=SEED, val=bool(TRAIN_ON_VAL))
+      else:
+        from examples.mlperf.dataloader import batch_load_llama3
+        return batch_load_llama3(GBS, SAMPLES, SEQLEN, BASEDIR, seed=SEED, val=bool(TRAIN_ON_VAL))
+
+  def get_eval_iter():
+    if getenv("FAKEDATA", 0):
+      return fake_data(EVAL_BS, 5760)
+    else:
+      if SMALL:
+        from examples.mlperf.dataloader import batch_load_llama3_small
+        return batch_load_llama3_small(EVAL_BS, 5760, SEQLEN, BASEDIR, val=True)
+      else:
+        from examples.mlperf.dataloader import batch_load_llama3
+        return batch_load_llama3(EVAL_BS, 5760, SEQLEN, BASEDIR, val=True)
+
+  iter = get_train_iter()
+  i, sequences_seen = resume_ckpt, 0
  for tokens in tqdm(iter, total=SAMPLES//GBS):
    t = time.perf_counter()
    GlobalCounters.reset()
    loss, lr = train_step(model, tokens, grad_acc)
    loss = loss.float().item()
-    # above as tqdm.write f-string
+
+    i += 1
+    sequences_seen += tokens.shape[0]
+
    tqdm.write(f"{loss:.4f} loss, {lr.item():.12f} LR, {GlobalCounters.mem_used / 1e9:.2f} GB used, {time.perf_counter()-t:.2f} s")
    if (fname:=getenv("LOSS_FILE", "")):
      with open(fname, "a") as f:
        f.write(f"{i} {loss:.4f} {lr.item():.12f} {GlobalCounters.mem_used / 1e9:.2f}\n")

-    if getenv("CKPT") and (i % 200 == 0 or i == 10):
+    if (ckpt_freq := getenv("CKPT")) and (i % ckpt_freq == 0 and (i != 1 or ckpt_freq == 1)):
      tqdm.write("saving checkpoint")
      if not os.path.exists(ckpt_dir := "./ckpts"): os.mkdir(ckpt_dir)
-      fn = f"{ckpt_dir}/{i}.safe"
+      fn = f"{ckpt_dir}/llama3_{i}.safe"
      safe_save(get_state_dict(model), fn)
-    i += 1
+
+      tqdm.write("saving optim checkpoint")
+      fn = f"{ckpt_dir}/llama3_{i}_optim.safe"
+      safe_save(get_state_dict(scheduler), fn)
+
+    if sequences_seen % EVAL_FREQ == 0 and (i != 1 or EVAL_FREQ == 1):
+      tqdm.write(f"evaluating after {sequences_seen} sequences")
+
+      # run eval
+      eval_losses = []
+      eval_iter = get_eval_iter()
+      tqdm.write(f"evaluating {5760//EVAL_BS} batches of {EVAL_BS} sequences")
+
+      for tokens in tqdm(eval_iter, total=5760//EVAL_BS):
+        eval_losses += eval_step(model, tokens).tolist()
+      log_perplexity = Tensor(eval_losses).mean().float().item()
+
+      tqdm.write(f"eval log perplexity: {log_perplexity:.4f}")
+
+      if log_perplexity < EVAL_TARGET:
+        tqdm.write(f"target achieved after {sequences_seen} sequences")
+        if getenv("CKPT"):
+          if not os.path.exists(ckpt_dir := "./ckpts"): os.mkdir(ckpt_dir)
+          fn = f"{ckpt_dir}/llama3.safe"
+          safe_save(get_state_dict(model), fn)
+        break
+
+def train_stable_diffusion():
+  from extra.models.unet import UNetModel
+  from examples.mlperf.dataloader import batch_load_train_stable_diffusion
+  from examples.mlperf.lr_schedulers import LambdaLR, LambdaLinearScheduler
+  from examples.mlperf.initializers import init_stable_diffusion
+  from examples.mlperf.helpers import get_training_state
+  import numpy as np
+
+  config = {}
+  GPUS               = config["GPUS"]                   = [f"{Device.DEFAULT}:{i}" for i in range(getenv("GPUS", 1))]
+  seed               = config["seed"]                   = getenv("SEED", 12345)
+  # ** hyperparameters **
+  BS                 = config["BS"]                     = getenv("BS", 1 * len(GPUS))
+  BASE_LR            = config["LEARNING_RATE"]          = getenv("LEARNING_RATE", 2.5e-7)
+  # https://github.com/mlcommons/training_policies/blob/cfa99da479b8d5931f7a3c67612d021dfb47510a/training_rules.adoc#benchmark_specific_rules
+  # "Checkpoint must be collected every 512,000 images. CEIL(512000 / global_batch_size) if 512000 is not divisible by GBS."
+  # NOTE: It's inferred that "steps" is the unit for the output of the CEIL formula, based on all other cases of CEIL in the rules
+  CKPT_STEP_INTERVAL = config["CKPT_STEP_INTERVAL"]     = getenv("CKPT_STEP_INTERVAL", math.ceil(512_000 / BS))
+  CKPTDIR            = config["CKPTDIR"]                = Path(getenv("CKPTDIR", "./checkpoints"))
+  DATADIR            = config["DATADIR"]                = Path(getenv("DATADIR", "./datasets"))
+  UNET_CKPTDIR       = config["UNET_CKPTDIR"]           = Path(getenv("UNET_CKPTDIR", "./checkpoints"))
+  TOTAL_CKPTS        = config["TOTAL_CKPTS"]            = getenv("TOTAL_CKPTS", 0)
+
+  print(f"training on {GPUS}")
+  lr = BS * BASE_LR
+  print(f"BS={BS}, BASE_LR={BASE_LR}, lr={lr}")
+  print(f"CKPT_STEP_INTERVAL = {CKPT_STEP_INTERVAL}")
+  for x in GPUS: Device[x]
+  if (WANDB := getenv("WANDB", "")):
+    import wandb
+    wandb.init(config=config, project="MLPerf-Stable-Diffusion")
+
+  Tensor.manual_seed(seed)  # seed for weight initialization
+  model, unet, sqrt_alphas_cumprod, sqrt_one_minus_alphas_cumprod = init_stable_diffusion("v2-mlperf-train", CKPTDIR / "sd" / "512-base-ema.ckpt", GPUS)
+
+  optimizer = AdamW(get_parameters(unet))
+  lambda_lr_callback = LambdaLinearScheduler(1000, 1.0, 1.0, 1e-06, 10000000000000).schedule
+  lr_scheduler = LambdaLR(optimizer, Tensor(lr, dtype=dtypes.float, device=optimizer.device), lambda_lr_callback)
+
+  @TinyJit
+  def train_step(mean:Tensor, logvar:Tensor, tokens:Tensor, unet:UNetModel, optimizer:LAMB, lr_scheduler:LambdaLR) -> Tensor:
+    optimizer.zero_grad()
+
+    timestep = Tensor.randint(BS, low=0, high=model.alphas_cumprod.shape[0], dtype=dtypes.int, device=GPUS[0])
+    latent_randn = Tensor.randn(*mean.shape, device=GPUS[0])
+    noise = Tensor.randn(*mean.shape, device=GPUS[0])
+    for t in (mean, logvar, tokens, timestep, latent_randn, noise):
+      t.shard_(GPUS, axis=0)
+
+    std = Tensor.exp(0.5 * logvar.clamp(-30.0, 20.0))
+    latent = (mean + std * latent_randn) * 0.18215
+
+    sqrt_alphas_cumprod_t = sqrt_alphas_cumprod[timestep].reshape(timestep.shape[0], 1, 1, 1)
+    sqrt_one_minus_alphas_cumprod_t = sqrt_one_minus_alphas_cumprod[timestep].reshape(timestep.shape[0], 1, 1, 1)
+    latent_with_noise = sqrt_alphas_cumprod_t * latent + sqrt_one_minus_alphas_cumprod_t * noise
+    v_true = sqrt_alphas_cumprod_t * noise - sqrt_one_minus_alphas_cumprod_t * latent
+
+    context = model.cond_stage_model.embed_tokens(tokens)
+
+    out = unet(latent_with_noise, timestep, context)
+    loss = ((out - v_true) ** 2).mean()
+    del mean, logvar, std, latent, noise, sqrt_alphas_cumprod_t, sqrt_one_minus_alphas_cumprod_t
+    del out, v_true, context, latent_randn, tokens, timestep
+    loss.backward()
+
+    optimizer.step()
+    lr_scheduler.step()
+    loss, out_lr = loss.detach().to("CPU"), optimizer.lr.to("CPU")
+    Tensor.realize(loss, out_lr)
+    return loss, out_lr
+    
+  # checkpointing takes ~9 minutes without this, and ~1 minute with this
+  @TinyJit
+  def ckpt_to_cpu():
+    ckpt = get_training_state(unet, optimizer, lr_scheduler)
+    # move to CPU first so more GPU bufs aren't created (can trigger OOM)
+    for k,v in ckpt.items(): ckpt[k] = v.detach().to("CPU")
+    Tensor.realize(*[v for v in ckpt.values()])
+    for k,v in ckpt.items(): ckpt[k] = v.cast(v.dtype.base).contiguous()
+    Tensor.realize(*[v for v in ckpt.values()])
+    return ckpt
+
+  # training loop
+  dl = batch_load_train_stable_diffusion(f'{DATADIR}/laion-400m/webdataset-moments-filtered/{{00000..00831}}.tar', BS)
+  # for tests
+  saved_checkpoints = []
+
+  train_start_time = time.perf_counter()
+  t0 = t6 = time.perf_counter()
+  for i, batch in enumerate(dl, start=1):
+    loop_time = time.perf_counter() - t0
+    t0 = time.perf_counter()
+    dl_time = t0 - t6
+    GlobalCounters.reset()
+
+    mean, logvar = np.split(np.concatenate(batch["npy"], axis=0), 2, axis=1)
+    mean, logvar = Tensor(mean, dtype=dtypes.float32, device="CPU"), Tensor(logvar, dtype=dtypes.float32, device="CPU")
+    tokens = []
+    for text in batch['txt']: tokens += model.cond_stage_model.tokenizer.encode(text, pad_with_zeros=True)
+    tokens = Tensor(tokens, dtype=dtypes.int32, device="CPU").reshape(-1, 77)
+
+    t1 = time.perf_counter()
+    loss, lr = train_step(mean, logvar, tokens, unet, optimizer, lr_scheduler)
+    loss_item, lr_item = loss.item(), lr.item()
+    t2 = time.perf_counter()
+
+    if i == 3:
+      for _ in range(3): ckpt_to_cpu() # do this at the beginning of run to prevent OOM surprises when checkpointing
+      print("BEAM COMPLETE", flush=True) # allows wrapper script to detect BEAM search completion and retry if it failed
+      
+    total_train_time = time.perf_counter() - train_start_time
+    if WANDB:
+      wandb.log({"train/loss": loss_item, "train/lr": lr_item, "train/loop_time_prev": loop_time, "train/dl_time": dl_time, "train/step": i,
+                    "train/GFLOPS": GlobalCounters.global_ops * 1e-9 / (t2-t1), "train/input_prep_time": t1-t0,
+                    "train/train_step_time": t2-t1, "train/total_time": total_train_time})
+
+      if i == 1 and wandb.run is not None:
+        with open(f"{UNET_CKPTDIR}/wandb_run_id_{wandb.run.id}", "w") as f:
+          f.write(f"wandb.run.id = {wandb.run.id}")
+
+    if i % CKPT_STEP_INTERVAL == 0:
+      # https://github.com/mlcommons/training_policies/blob/cfa99da479b8d5931f7a3c67612d021dfb47510a/training_rules.adoc#benchmark_specific_rules
+      # "evaluation is done offline, the time is not counted towards the submission time."
+      fn = f"{UNET_CKPTDIR}/{i}.safetensors"
+      print(f"saving unet checkpoint at {fn}")
+      saved_checkpoints.append(fn)
+      safe_save({k.replace("model.", ""):v for k,v in ckpt_to_cpu().items() if k.startswith("model.")}, fn)
+      if TOTAL_CKPTS and i == TOTAL_CKPTS * CKPT_STEP_INTERVAL:
+        print(f"ending run after {i} steps ({TOTAL_CKPTS} checkpoints collected)")
+        return saved_checkpoints
+
+    t3 = time.perf_counter()
+    print(f"""step {i}: {GlobalCounters.global_ops * 1e-9 / (t2-t1):9.2f} GFLOPS, mem_used: {GlobalCounters.mem_used / 1e9:.2f} GB,
+  loop_time_prev: {loop_time:.2f}, dl_time: {dl_time:.2f}, input_prep_time: {t1-t0:.2f}, train_step_time: {t2-t1:.2f},
+  t3-t2: {t3-t2:.4f}, loss:{loss_item:.5f}, lr:{lr_item:.3e}, total_train_time:{total_train_time:.2f}
+  """)
+    t6 = time.perf_counter()

 if __name__ == "__main__":
  multiprocessing.set_start_method('spawn')
@@ -1414,7 +1638,7 @@ if __name__ == "__main__":
  else: bench_log_manager = contextlib.nullcontext()

  with Tensor.train():
-    for m in getenv("MODEL", "resnet,retinanet,unet3d,rnnt,bert,maskrcnn").split(","):
+    for m in getenv("MODEL", "resnet,retinanet,unet3d,rnnt,bert,maskrcnn,stable_diffusion").split(","):
      nm = f"train_{m}"
      if nm in globals():
        print(f"training {m}")
@@ -0,0 +1,57 @@
+#!/usr/bin/env bash
+# adapted from https://github.com/mlcommons/training/blob/4bdf5c8ed218ad76565a2ba1ac27c919ccc6d689/stable_diffusion/README.md
+
+# setup dirs
+
+DATA=/raid/datasets/stable_diffusion
+
+LAION=$DATA/laion-400m/webdataset-moments-filtered 
+COCO=$DATA/coco2014
+mkdir -p $LAION $COCO
+
+CKPT=/raid/weights/stable_diffusion
+mkdir -p $CKPT/clip $CKPT/sd $CKPT/inception
+
+# download data
+
+# if rclone isn't installed system-wide / in your PATH, put the executable path in quotes below
+#RCLONE=""
+RCLONE="rclone"
+
+## VAE-encoded image latents, from 6.1M image subset of laion-400m
+## about 1 TB for whole download
+$RCLONE config create mlc-training s3 provider=Cloudflare access_key_id=76ea42eadb867e854061a1806220ee1e secret_access_key=a53625c4d45e3ca8ac0df8a353ea3a41ffc3292aa25259addd8b7dc5a6ce2936 endpoint=c2686074cb2caf5cbaf6d134bdba8b47.r2.cloudflarestorage.com
+$RCLONE copy mlc-training:mlcommons-training-wg-public/stable_diffusion/datasets/laion-400m/moments-webdataset-filtered/ ${LAION} --include="*.tar" -P
+$RCLONE copy mlc-training:mlcommons-training-wg-public/stable_diffusion/datasets/laion-400m/moments-webdataset-filtered/sha512sums.txt ${LAION} -P
+cd $LAION && grep -E '\.tar$' sha512sums.txt | sha512sum -c --quiet - && \
+  echo "All .tar files verified" || { echo "Checksum failure when validating downloaded Laion moments"; exit 1; }
+
+## prompts and FID statistics from 30k image subset of coco2014
+## 33 MB
+$RCLONE config create mlc-training s3 provider=Cloudflare access_key_id=76ea42eadb867e854061a1806220ee1e secret_access_key=a53625c4d45e3ca8ac0df8a353ea3a41ffc3292aa25259addd8b7dc5a6ce2936 endpoint=c2686074cb2caf5cbaf6d134bdba8b47.r2.cloudflarestorage.com
+$RCLONE copy mlc-training:mlcommons-training-wg-public/stable_diffusion/datasets/coco2014/val2014_30k.tsv ${COCO} -P
+
+$RCLONE config create mlc-training s3 provider=Cloudflare access_key_id=76ea42eadb867e854061a1806220ee1e secret_access_key=a53625c4d45e3ca8ac0df8a353ea3a41ffc3292aa25259addd8b7dc5a6ce2936 endpoint=c2686074cb2caf5cbaf6d134bdba8b47.r2.cloudflarestorage.com
+$RCLONE copy mlc-training:mlcommons-training-wg-public/stable_diffusion/datasets/coco2014/val2014_30k_stats.npz ${COCO} -P
+
+# download checkpoints
+
+## clip (needed for text and vision encoders for validation)
+CLIP_WEIGHTS_URL="https://huggingface.co/laion/CLIP-ViT-H-14-laion2B-s32B-b79K/resolve/main/open_clip_pytorch_model.bin"
+CLIP_WEIGHTS_SHA256="9a78ef8e8c73fd0df621682e7a8e8eb36c6916cb3c16b291a082ecd52ab79cc4"
+CLIP_CONFIG_URL="https://huggingface.co/laion/CLIP-ViT-H-14-laion2B-s32B-b79K/raw/main/open_clip_config.json"
+wget -N -P ${CKPT}/clip ${CLIP_WEIGHTS_URL}
+wget -N -P ${CKPT}/clip ${CLIP_CONFIG_URL}
+echo "${CLIP_WEIGHTS_SHA256}  ${CKPT}/clip/open_clip_pytorch_model.bin"                    | sha256sum -c
+
+## sd (needed for latent->image decoder for validation, also has clip text encoder for training)
+SD_WEIGHTS_URL='https://huggingface.co/stabilityai/stable-diffusion-2-base/resolve/main/512-base-ema.ckpt'
+SD_WEIGHTS_SHA256="d635794c1fedfdfa261e065370bea59c651fc9bfa65dc6d67ad29e11869a1824"
+wget -N -P ${CKPT}/sd ${SD_WEIGHTS_URL}
+echo "${SD_WEIGHTS_SHA256}  ${CKPT}/sd/512-base-ema.ckpt"                    | sha256sum -c
+
+## inception (needed for validation)
+FID_WEIGHTS_URL='https://github.com/mseitzer/pytorch-fid/releases/download/fid_weights/pt_inception-2015-12-05-6726825d.pth'
+FID_WEIGHTS_SHA1="bd836944fd6db519dfd8d924aa457f5b3c8357ff"
+wget -N -P ${CKPT}/inception ${FID_WEIGHTS_URL}
+echo "${FID_WEIGHTS_SHA1}  ${CKPT}/inception/pt_inception-2015-12-05-6726825d.pth"                    | sha1sum -c
@@ -0,0 +1,72 @@
+#!/usr/bin/env bash
+
+DATETIME=${2:-$(date "+%m%d%H%M")}
+LOGFILE="${HOME}/logs/sd_mi300x_${DATETIME}.log"
+# UNET_CKPTDIR must be set: training saves checkpoints to this path, then a separate eval process scans this path to know which checkpoints to eval
+export UNET_CKPTDIR="${HOME}/stable_diffusion/training_checkpoints/${DATETIME}"
+mkdir -p "${HOME}/logs" "$UNET_CKPTDIR"
+
+# run this script in isolation when using the --bg flag
+if [[ "${1:-}" == "--bg" ]]; then
+  echo "logging output to $LOGFILE"
+  echo "saving UNet checkpoints to $UNET_CKPTDIR"
+  script_path="$(readlink -f "${BASH_SOURCE[0]}")"
+  nohup bash "$script_path" run "$DATETIME" >"$LOGFILE" 2>&1 & disown $!
+  exit 0
+fi
+
+# venv management
+if [[ -d .venv-sd-mlperf ]]; then
+  . .venv-sd-mlperf/bin/activate
+else
+  python3 -m venv .venv-sd-mlperf && . .venv-sd-mlperf/bin/activate
+  pip install --index-url https://download.pytorch.org/whl/cpu torch && pip install tqdm numpy ftfy regex pillow scipy wandb webdataset
+fi
+pip list
+apt list --installed | grep amdgpu
+rocm-smi --version
+modinfo amdgpu | grep version
+
+export BEAM=2 BEAM_UOPS_MAX=8000 BEAM_UPCAST_MAX=256 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=5 IGNORE_JIT_FIRST_BEAM=1 HCQDEV_WAIT_TIMEOUT_MS=300000
+export AMD_LLVM=0 # bf16 seems to require this
+export DATADIR="/raid/datasets/stable_diffusion"
+export CKPTDIR="/raid/weights/stable_diffusion"
+export EVAL_CKPT_DIR=$UNET_CKPTDIR
+export MODEL="stable_diffusion" PYTHONPATH="."
+export GPUS=8 BS=304
+export CONTEXT_BS=816 DENOISE_BS=600 DECODE_BS=384 INCEPTION_BS=560 CLIP_BS=240
+export WANDB=1
+export PARALLEL=4
+export PYTHONUNBUFFERED=1
+sudo rocm-smi -d 0 1 2 3 4 5 6 7 --setperfdeterminism 1500 || exit 1
+
+# Retry BEAM search if script fails before BEAM COMPLETE is printed, but don't retry after that
+run_retry(){ local try=0 max=5 code tmp py pgid kids
+  while :; do
+    tmp=$(mktemp)
+    setsid bash -c 'exec env "$@"' _ "$@" > >(tee -a "$LOGFILE" | tee "$tmp") 2>&1 &
+    py=$!; pgid=$(ps -o pgid= -p "$py" | tr -d ' ')
+    wait "$py"; code=$?
+    [[ -n "$pgid" ]] && { kill -TERM -"$pgid" 2>/dev/null; sleep 1; kill -KILL -"$pgid" 2>/dev/null; }
+    kids=$(pgrep -P "$py" || true)
+    while [[ -n "$kids" ]]; do
+      kill -TERM $kids 2>/dev/null; sleep 0.5
+      kids=$(for k in $kids; do pgrep -P "$k" || true; done)
+    done
+    grep -q 'BEAM COMPLETE' "$tmp" && { rm -f "$tmp"; return 1; }
+    rm -f "$tmp"
+    ((code==0)) && return 0
+    ((try>=max)) && return 2
+    ((try++)); sleep 90; echo "try = ${try}"
+  done
+}
+
+# Power limiting to 400W is only needed if GPUs fall out of sync (causing 2.2x increased train time) at higher power, which has been observed at 450W
+sudo rocm-smi -d 0 1 2 3 4 5 6 7 --setpoweroverdrive 750 && \
+run_retry TOTAL_CKPTS=7 python3 examples/mlperf/model_train.py; (( $? == 2 )) && { echo "training failed before BEAM completion"; exit 2; }
+sleep 90
+
+run_retry EVAL_SAMPLES=600 python3 examples/mlperf/model_eval.py; (( $? == 2 )) && { echo "eval failed before BEAM completion"; exit 2; }
+# Checkpoints will be evaluated in reverse chronological order, even if above training crashed early
+# STOP_IF_CONVERGED=1: Stop the eval after the first time convergence is detected; no more checkpoints will be evaluated after that.
+STOP_IF_CONVERGED=1 python3 examples/mlperf/model_eval.py
@@ -10,7 +10,7 @@ from tinygrad.helpers import DEBUG, getenv
 from tinygrad.engine.realize import CompiledRunner

 import onnx
-from tinygrad.frontend.onnx import OnnxRunner
+from tinygrad.nn.onnx import OnnxRunner

 OPENPILOT_MODEL = sys.argv[1] if len(sys.argv) > 1 else "https://github.com/commaai/openpilot/raw/v0.9.7/selfdrive/modeld/models/supercombo.onnx"
 OUTPUT = sys.argv[2] if len(sys.argv) > 2 else "/tmp/openpilot.pkl"
@@ -77,13 +77,20 @@ def test_vs_compile(run, new_inputs, test_val=None):
            **{k:Tensor(v, device="NPY").realize() for k,v in new_inputs_numpy.items() if 'img' not in k}}

  # run 20 times
+  step_times = []
  for _ in range(20):
    st = time.perf_counter()
    out = run(**inputs)
    mt = time.perf_counter()
    val = out.numpy()
    et = time.perf_counter()
-    print(f"enqueue {(mt-st)*1e3:6.2f} ms -- total run {(et-st)*1e3:6.2f} ms")
+    step_times.append((et-st)*1e3)
+    print(f"enqueue {(mt-st)*1e3:6.2f} ms -- total run {step_times[-1]:6.2f} ms")
+
+  if (assert_time:=getenv("ASSERT_MIN_STEP_TIME")):
+    min_time = min(step_times)
+    assert min_time < assert_time, f"Speed regression, expected min step time of < {assert_time} ms but took: {min_time} ms"
+
  print(out, val.shape, val.dtype)
  if test_val is not None: np.testing.assert_equal(test_val, val)
  print("**** test done ****")
@@ -1,12 +1,12 @@
 import sys
 from tinygrad import Tensor, fetch, GlobalCounters, dtypes
 from tinygrad.uop.ops import UOp
-from tinygrad.frontend.onnx import OnnxRunner
-from tinygrad.schedule.kernelize import get_kernelize_map
+from tinygrad.nn.onnx import OnnxRunner
+from tinygrad.schedule.rangeify import get_rangeify_map
 from tinygrad.engine.schedule import create_schedule_with_vars
 from tinygrad.engine.realize import run_schedule

-# NOLOCALS=1 GPU=1 IMAGE=2 FLOAT16=1 VIZ=1 DEBUG=2 python3 examples/openpilot/compile4.py
+# NOLOCALS=1 CL=1 IMAGE=2 FLOAT16=1 VIZ=1 DEBUG=2 python3 examples/openpilot/compile4.py

 OPENPILOT_MODEL = sys.argv[1] if len(sys.argv) > 1 else "https://github.com/commaai/openpilot/raw/v0.9.7/selfdrive/modeld/models/supercombo.onnx"
 OUTPUT = sys.argv[2] if len(sys.argv) > 2 else "/tmp/openpilot.pkl"
@@ -33,7 +33,7 @@ if __name__ == "__main__":
        if not in_target_path[s]:
          independent_set[s] = None
  independent = UOp.sink(*independent_set.keys())
-  kernelized = get_kernelize_map(independent)
+  kernelized = get_rangeify_map(independent)
  independent = independent.substitute(kernelized)
  schedule, var_vals = create_schedule_with_vars(independent)
  run_schedule(schedule)
@@ -27,7 +27,7 @@ class Model(nn.Module):

 if __name__ == "__main__":
  if getenv("TINY_BACKEND"):
-    import tinygrad.frontend.torch  # noqa: F401
+    import tinygrad.nn.torch  # noqa: F401
    device = torch.device("tiny")
  else:
    device = torch.device({"METAL":"mps","NV":"cuda"}.get(Device.DEFAULT, "cpu"))
@@ -8,7 +8,7 @@ from typing import Dict, Union

 from extra.models.llama import Transformer, convert_from_huggingface, fix_bf16
 from examples.llama3 import load
-from tinygrad import nn, Tensor
+from tinygrad import nn, Tensor, Device
 from tinygrad.helpers import fetch, colored, GlobalCounters, Timing, DEBUG
 from tinygrad.nn.state import load_state_dict, get_parameters

@@ -80,7 +80,7 @@ if __name__ == "__main__":
    st = GlobalCounters.time_sum_s
    next_tok = Tensor([toks[start_pos:]]) if tok_tensor is None or (len(toks)-start_pos) > 1 else tok_tensor.reshape(1, 1)
    with Timing("total ", enabled=args.timing, on_exit=lambda x: f", {1e9/x:.2f} tok/s, {GlobalCounters.global_mem/x:.2f} GB/s, param {param_bytes/x:.2f} GB/s"):
-      with Timing("enqueue in ", on_exit=(lambda et: (f", {(GlobalCounters.time_sum_s-st)*1e3:.2f} ms on GPU" if DEBUG>=2 else "") +
+      with Timing("enqueue in ", on_exit=(lambda et: (f", {(GlobalCounters.time_sum_s-st)*1e3:.2f} ms on {Device.DEFAULT}" if DEBUG>=2 else "") +
                  f", {GlobalCounters.global_ops*1e-9:.2f} GOPS, {GlobalCounters.global_mem*1e-9:.2f} GB" +
                  (f", {GlobalCounters.global_mem*1e-9/(GlobalCounters.time_sum_s-st):.2f} GB/s, param {param_bytes*1e-9/(GlobalCounters.time_sum_s-st):.2f} GB/s" if DEBUG>=2 else "")) if DEBUG else None, enabled=args.timing):
        tok_tensor = transformer(next_tok, start_pos, args.temperature)
@@ -6,7 +6,7 @@
 from tinygrad import Tensor, TinyJit, dtypes, GlobalCounters
 from tinygrad.nn import Conv2d, GroupNorm
 from tinygrad.nn.state import safe_load, load_state_dict
-from tinygrad.helpers import fetch, trange, colored, Timing
+from tinygrad.helpers import fetch, trange, colored, Timing, getenv
 from extra.models.clip import Embedder, FrozenClosedClipEmbedder, FrozenOpenClipEmbedder
 from extra.models.unet import UNetModel, Upsample, Downsample, timestep_embedding
 from extra.bench_log import BenchEvent, WallTimeEvent
@@ -14,7 +14,7 @@ from examples.stable_diffusion import ResnetBlock, Mid
 import numpy as np

 from typing import Dict, List, Callable, Optional, Any, Set, Tuple, Union, Type
-import argparse, tempfile
+import argparse, tempfile, time
 from abc import ABC, abstractmethod
 from pathlib import Path
 from PIL import Image
@@ -342,11 +342,13 @@ class DPMPP2MSampler:
    sigmas = self.discretization(num_steps).to(x.device)
    x *= Tensor.sqrt(1.0 + sigmas[0] ** 2.0)
    num_sigmas = len(sigmas)
+    step_times = []

    old_denoised = None
    for i in trange(num_sigmas - 1):
      with Timing("step in ", enabled=timing, on_exit=lambda _: f", using {GlobalCounters.mem_used/1e9:.2f} GB"):
        GlobalCounters.reset()
+        st = time.perf_counter_ns()
        with WallTimeEvent(BenchEvent.STEP):
          x, old_denoised = self.sampler_step(
            old_denoised=old_denoised,
@@ -358,8 +360,13 @@ class DPMPP2MSampler:
            c=c,
            uc=uc,
          )
+          step_times.append(t:=(time.perf_counter_ns() - st)*1e-6)
          x.realize(old_denoised)

+    if (assert_time:=getenv("ASSERT_MIN_STEP_TIME")):
+      min_time = min(step_times)
+      assert min_time < assert_time, f"Speed regression, expected min step time of < {assert_time} ms but took: {min_time} ms"
+
    return x


@@ -430,8 +437,8 @@ if __name__ == "__main__":
    im.show()

  # validation!
-  if args.prompt == default_prompt and args.steps == 10 and args.seed == 0 and args.guidance == 6.0 and args.width == args.height == 1024 \
-    and not args.weights:
+  is_default = args.prompt == default_prompt and args.steps == 10 and args.seed == 0 and args.guidance == 6.0 and args.width == args.height == 1024
+  if is_default and not args.weights and not args.fakeweights:
    ref_image = Tensor(np.array(Image.open(Path(__file__).parent / "sdxl_seed0.png")))
    distance = (((x.cast(dtypes.float) - ref_image.cast(dtypes.float)) / ref_image.max())**2).mean().item()
    assert distance < 4e-3, colored(f"validation failed with {distance=}", "red")
@@ -2,18 +2,20 @@
 # https://github.com/ekagra-ranjan/huggingface-blog/blob/main/stable_diffusion.md
 import tempfile
 from pathlib import Path
-import argparse
+import argparse, time
 from collections import namedtuple
 from typing import Dict, Any

 from PIL import Image
 import numpy as np
 from tinygrad import Device, GlobalCounters, dtypes, Tensor, TinyJit
-from tinygrad.helpers import Timing, Context, getenv, fetch, colored, tqdm
+from tinygrad.helpers import Timing, Context, getenv, fetch, colored, tqdm, flatten
 from tinygrad.nn import Conv2d, GroupNorm
 from tinygrad.nn.state import torch_load, load_state_dict, get_state_dict
-from extra.models.clip import Closed, Tokenizer
+from extra.models.clip import Closed, Tokenizer, FrozenOpenClipEmbedder
+from extra.models import unet, clip
 from extra.models.unet import UNetModel
+from examples.mlperf.initializers import AutocastLinear, AutocastConv2d, AutocastGroupNorm, AutocastLayerNorm, zero_module, attn_f32_softmax, gelu_erf
 from extra.bench_log import BenchEvent, WallTimeEvent

 class AttnBlock:
@@ -154,12 +156,46 @@ unet_params: Dict[str,Any] = {
  "use_linear": False,
 }

+mlperf_params: Dict[str,Any] = {"adm_in_ch": None, "in_ch": 4, "out_ch": 4, "model_ch": 320, "attention_resolutions": [4, 2, 1], "num_res_blocks": 2,
+                                "channel_mult": [1, 2, 4, 4], "d_head": 64, "transformer_depth": [1, 1, 1, 1], "ctx_dim": 1024, "use_linear": True,
+                                "num_groups":16, "st_norm_eps":1e-6}
+
 class StableDiffusion:
-  def __init__(self):
+  def __init__(self, version:str|None=None, pretrained:str|None=None):
    self.alphas_cumprod = get_alphas_cumprod()
-    self.model = namedtuple("DiffusionModel", ["diffusion_model"])(diffusion_model = UNetModel(**unet_params))
-    self.first_stage_model = AutoencoderKL()
-    self.cond_stage_model = namedtuple("CondStageModel", ["transformer"])(transformer = namedtuple("Transformer", ["text_model"])(text_model = Closed.ClipTextTransformer()))
+    if version != "v2-mlperf-train":
+      self.first_stage_model = AutoencoderKL() # only needed for decoding generated latents to images; not needed in mlperf training from preprocessed moments
+
+    if not version:
+      self.cond_stage_model = namedtuple("CondStageModel", ["transformer"])(transformer = namedtuple("Transformer", ["text_model"])(text_model = Closed.ClipTextTransformer()))
+      unet_init_params = unet_params
+    elif version in {"v2-mlperf-train", "v2-mlperf-eval"}:
+      unet_init_params = mlperf_params
+      clip.gelu = gelu_erf
+      self.cond_stage_model = FrozenOpenClipEmbedder(**{"dims": 1024, "n_heads": 16, "layers": 24, "return_pooled": False, "ln_penultimate": True,
+                                                        "clip_tokenizer_version": "sd_mlperf_v5_0"})
+      unet.Linear, unet.Conv2d, unet.GroupNorm, unet.LayerNorm = AutocastLinear, AutocastConv2d, AutocastGroupNorm, AutocastLayerNorm
+      unet.attention, unet.gelu, unet.mixed_precision_dtype = attn_f32_softmax, gelu_erf, dtypes.bfloat16
+      if pretrained:
+        print("loading text encoder")
+        weights: dict[str,Tensor] = {k.replace("cond_stage_model.", "", 1):v for k,v in torch_load(pretrained)["state_dict"].items() if k.startswith("cond_stage_model.")}
+        weights["model.attn_mask"] = Tensor.full((77, 77), fill_value=float("-inf")).triu(1)
+        load_state_dict(self.cond_stage_model, weights)
+        # only the eval model needs the decoder
+        if version == "v2-mlperf-eval":
+          print("loading image latent encoder")
+          weights = {k.replace("first_stage_model.", "", 1):v for k,v in torch_load(pretrained)["state_dict"].items() if k.startswith("first_stage_model.")}
+          load_state_dict(self.first_stage_model, weights)
+
+    self.model = namedtuple("DiffusionModel", ["diffusion_model"])(diffusion_model = UNetModel(**unet_init_params))
+    if version == "v2-mlperf-train":
+      # the mlperf reference inits certain weights as zeroes
+      for bb in flatten(self.model.diffusion_model.input_blocks) + self.model.diffusion_model.middle_block + flatten(self.model.diffusion_model.output_blocks):
+        if isinstance(bb, unet.ResBlock):
+          zero_module(bb.out_layers[3])
+        elif isinstance(bb, unet.SpatialTransformer):
+          zero_module(bb.proj_out)
+      zero_module(self.model.diffusion_model.out[2])

  def get_x_prev_and_pred_x0(self, x, e_t, a_t, a_prev):
    temperature = 1
@@ -233,12 +269,14 @@ if __name__ == "__main__":

  # load in weights
  with WallTimeEvent(BenchEvent.LOAD_WEIGHTS):
-    load_state_dict(model, torch_load(fetch('https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/resolve/main/sd-v1-4.ckpt', 'sd-v1-4.ckpt'))['state_dict'], strict=False)
+    load_state_dict(model, torch_load(fetch('https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/resolve/main/sd-v1-4.ckpt', 'sd-v1-4.ckpt'))['state_dict'], verbose=False, strict=False, realize=False)

    if args.fp16:
      for k,v in get_state_dict(model).items():
        if k.startswith("model"):
-          v.replace(v.cast(dtypes.float16).realize())
+          v.replace(v.cast(dtypes.float16))
+
+    Tensor.realize(*get_state_dict(model).values())

  # run through CLIP to get context
  tokenizer = Tokenizer.ClipTokenizer()
@@ -266,17 +304,23 @@ if __name__ == "__main__":
  def run(model, *x): return model(*x).realize()

  # this is diffusion
+  step_times = []
  with Context(BEAM=getenv("LATEBEAM")):
    for index, timestep in (t:=tqdm(list(enumerate(timesteps))[::-1])):
      GlobalCounters.reset()
+      st = time.perf_counter_ns()
      t.set_description("%3d %3d" % (index, timestep))
      with Timing("step in ", enabled=args.timing, on_exit=lambda _: f", using {GlobalCounters.mem_used/1e9:.2f} GB"):
        with WallTimeEvent(BenchEvent.STEP):
          tid = Tensor([index])
          latent = run(model, unconditional_context, context, latent, Tensor([timestep]), alphas[tid], alphas_prev[tid], Tensor([args.guidance]))
          if args.timing: Device[Device.DEFAULT].synchronize()
+      step_times.append((time.perf_counter_ns() - st)*1e-6)
    del run

+  if (assert_time:=getenv("ASSERT_MIN_STEP_TIME")):
+    min_time = min(step_times)
+    assert min_time < assert_time, f"Speed regression, expected min step time of < {assert_time} ms but took: {min_time} ms"
  # upsample latent space to image with autoencoder
  x = model.decode(latent)
  print(x.shape)
@@ -32,7 +32,7 @@ if __name__ == "__main__":

  lr = 5e-3
  transform = ComposeTransforms([
-    lambda x: [Image.fromarray(xx, mode='L').resize((64, 64)) for xx in x],
+    lambda x: [Image.fromarray(xx).resize((64, 64)) for xx in x],
    lambda x: np.stack([np.asarray(xx) for xx in x], 0),
    lambda x: x / 255.0,
    lambda x: np.tile(np.expand_dims(x, 1), (1, 3, 1, 1)).astype(np.float32),
@@ -109,7 +109,7 @@ class TextDecoder:

  def forward(self, x:Tensor, pos:Union[Variable, Literal[0]], encoded_audio:Tensor):
    seqlen = x.shape[-1]
-    x = self.token_embedding(x) + self.positional_embedding.shrink(((pos, pos+seqlen), None, None))
+    x = self.token_embedding(x) + self.positional_embedding.shrink(((pos, pos+seqlen), None))
    for block in self.blocks: x = block(x, xa=encoded_audio, mask=self.mask, len=pos)
    return self.output_tok(x)

@@ -2,7 +2,7 @@
 import os
 from ultralytics import YOLO
 from pathlib import Path
-from tinygrad.frontend.onnx import OnnxRunner
+from tinygrad.nn.onnx import OnnxRunner
 from extra.onnx_helpers import get_example_inputs

 os.chdir("/tmp")
@@ -1,6 +1,16 @@
-import re, ctypes, sys
+import re, ctypes, sys, importlib

-from tinygrad.runtime.autogen.am import am, mp_11_0, mp_13_0_0, nbio_4_3_0, mmhub_3_0_0, gc_11_0_0, osssys_6_0_0
+from tinygrad.runtime.support.am.amdev import AMDev, AMRegister
+class AMDFake(AMDev):
+  def __init__(self, devfmt, vram, doorbell, mmio, dma_regions=None):
+    self.devfmt, self.vram, self.doorbell64, self.mmio, self.dma_regions = devfmt, vram, doorbell, mmio, dma_regions
+    self._run_discovery()
+    self._build_regs()
+
+amdev = importlib.import_module("tinygrad.runtime.support.am.amdev")
+amdev.AMDev = AMDFake
+
+from tinygrad.runtime.ops_amd import PCIIface

 def parse_amdgpu_logs(log_content, register_names=None):
  register_map = register_names
@@ -23,16 +33,11 @@ def parse_amdgpu_logs(log_content, register_names=None):
  return processed_log

 def main():
-  regs_offset = {13: {0: [3072, 37784576]}, 28: {0: [93184, 37754880], 1: [201327616, 201461760], 2: [209716224, 209850368], 3: [218104832, 218238976], 4: [226493440, 226627584], 5: [234882048, 235016192], 6: [243270656, 243404800]}, 21: {0: [28672, 12582912, 37795840, 130023424, 306184192], 1: [201326592, 201463808, 201465856, 204210176, 204472320], 2: [209715200, 209852416, 209854464, 212598784, 212860928], 3: [218103808, 218241024, 218243072, 220987392, 221249536], 4: [226492416, 226629632, 226631680, 229376000, 229638144], 5: [234881024, 235018240, 235020288, 237764608, 238026752], 6: [243269632, 243406848, 243408896, 246153216, 246415360]}, 22: {0: [18, 192, 13504, 36864, 37764096]}, 1: {0: [4704, 40960, 114688, 37760000]}, 2: {0: [3872, 37790720]}, 11: {0: [70656, 38103040]}, 12: {0: [106496, 37783552]}, 15: {0: [90112, 14417920, 14680064, 14942208, 38009856]}, 16: {0: [90112, 14417920, 14680064, 14942208, 38009856]}, 14: {0: [0, 20, 3360, 66560, 37859328, 67371008]}, 26: {0: [0, 20, 3360, 66560, 37859328, 67371008]}, 23: {0: [4256, 37789696]}, 33: {0: [0, 20, 3360, 66560, 37859328, 67371008]}, 25: {0: []}, 3: {0: [4704, 40960, 114688, 37760000]}, 4: {0: [4704, 40960, 114688, 37760000]}, 24: {0: [92160, 92672, 37752832, 54788096]}, 27: {0: [91648, 37751808], 1: [201339904, 201458176], 2: [209728512, 209846784], 3: [218117120, 218235392], 4: [226505728, 226624000], 5: [234894336, 235012608], 6: [243282944, 243401216]}, 29: {0: [201342976, 201344000, 205520896, 205537280], 1: [209731584, 209732608, 213909504, 213925888], 2: [218120192, 218121216, 222298112, 222314496], 3: [226508800, 226509824, 230686720, 230703104], 4: [234897408, 234898432, 239075328, 239091712], 5: [243286016, 243287040, 247463936, 247480320]}, 17: {0: [30720, 32256], 1: [31488, 73728]}}
-
  reg_names = {}
-  def _prepare_registers(modules):
-    for base, m in modules:
-      for k, regval in m.__dict__.items():
-        if k.startswith("reg") and not k.endswith("_BASE_IDX") and (base_idx:=getattr(m, f"{k}_BASE_IDX", None)) is not None:
-          reg_names[regs_offset[am.__dict__.get(f"{base}_HWIP")][0][base_idx] + regval] = k
-
-  _prepare_registers([("MP0", mp_13_0_0), ("NBIO", nbio_4_3_0), ("MMHUB", mmhub_3_0_0), ("GC", gc_11_0_0), ("OSSSYS", osssys_6_0_0)])
+  dev = PCIIface(None, 0)
+  for x, y in dev.dev_impl.__dict__.items():
+    if isinstance(y, AMRegister):
+      for inst, addr in y.addr.items(): reg_names[addr] = f"{x}, xcc={inst}"

  with open(sys.argv[1], 'r') as f:
    log_content = log_content_them = f.read()
@@ -1,7 +1,7 @@
 # copying the kernels from https://github.com/microsoft/ArchProbe into Python
 import numpy as np
 import pickle
-from tinygrad.runtime.ops_gpu import CLProgram, CLBuffer
+from tinygrad.runtime.ops_cl import CLProgram, CLBuffer
 from tinygrad import dtypes
 from tqdm import trange, tqdm
 from matplotlib import pyplot as plt
@@ -4,7 +4,7 @@ from tinygrad import dtypes
 from tinygrad.codegen.assembly import AssemblyCodegen, Register
 from tinygrad.codegen.opt.kernel import Ops
 from tinygrad.uop.ops import BinaryOps, UnaryOps, TernaryOps
-from tinygrad.runtime.ops_gpu import ROCM_LLVM_PATH
+from tinygrad.runtime.ops_cl import ROCM_LLVM_PATH

 # ugh, is this really needed?
 from extra.helpers import enable_early_exec
@@ -5,7 +5,7 @@ from tinygrad.helpers import colored
 from extra.helpers import enable_early_exec
 early_exec = enable_early_exec()

-from tinygrad.runtime.ops_gpu import CLProgram, CLBuffer, ROCM_LLVM_PATH
+from tinygrad.runtime.ops_cl import CLProgram, CLBuffer, ROCM_LLVM_PATH

 ENABLE_NON_ASM = False

@@ -10,13 +10,13 @@ from tinygrad.renderer.cstyle import ClangRenderer
 render_dtype = ClangRenderer().render_dtype

 class ClangGraph(GraphRunner):
-  def __init__(self, jit_cache: List[ExecItem], input_rawbuffers: List[Buffer], var_vals: Dict[Variable, int]):
+  def __init__(self, jit_cache: List[ExecItem], input_rawbuffers: List[Buffer], var_vals: Dict[str, int]):
    super().__init__(jit_cache, input_rawbuffers, var_vals)
    if not all(isinstance(ji.prg, CompiledRunner) for ji in jit_cache): raise GraphException

    prgs = '\n'.join(dedup([cast(CompiledRunner, ji.prg).p.src for ji in jit_cache]))
    args = [f"{render_dtype(x.dtype)}* arg{i}" for i,x in enumerate(input_rawbuffers)]
-    args += sorted([f"int {v.expr}" for v in var_vals])
+    args += sorted([f"int {v}" for v in var_vals])
    code = ["void batched("+','.join(args)+") {"]
    for ji in jit_cache:
      args = []
@@ -34,6 +34,6 @@ class ClangGraph(GraphRunner):
    assert compiler is not None
    self._prg = ClangProgram("batched", compiler.compile(prgs+"\n"+"\n".join(code))) # no point in caching the pointers

-  def __call__(self, rawbufs: List[Buffer], var_vals: Dict[Variable, int], wait=False):
+  def __call__(self, rawbufs: List[Buffer], var_vals: Dict[str, int], wait=False):
    return cpu_time_execution(
-    lambda: self._prg(*[x._buf for x in rawbufs], *[x[1] for x in sorted(var_vals.items(), key=lambda x: x[0].expr)]), enable=wait)
+    lambda: self._prg(*[x._buf for x in rawbufs], *[x[1] for x in sorted(var_vals.items(), key=lambda x: x[0])]), enable=wait)
@@ -26,7 +26,7 @@ class VirtAQLQueue(AQLQueue):
    self.available_packet_slots -= 1

 class HSAGraph(MultiGraphRunner):
-  def __init__(self, jit_cache: List[ExecItem], input_rawbuffers: List[Buffer], var_vals: Dict[Variable, int]):
+  def __init__(self, jit_cache: List[ExecItem], input_rawbuffers: List[Buffer], var_vals: Dict[str, int]):
    super().__init__(jit_cache, input_rawbuffers, var_vals)

    # Check all jit items are compatible.
@@ -53,7 +53,7 @@ class HSAGraph(MultiGraphRunner):
      self.ji_kargs_structs[j] = ji.prg._prg.args_struct_t.from_address(kernargs_ptrs[ji.prg.dev])
      kernargs_ptrs[ji.prg.dev] += round_up(ctypes.sizeof(ji.prg._prg.args_struct_t), 16)
      for i in range(len(ji.bufs)): self.ji_kargs_structs[j].__setattr__(f'f{i}', cast(Buffer, ji.bufs[i])._buf)
-      for i in range(len(ji.prg.p.vars)): self.ji_kargs_structs[j].__setattr__(f'v{i}', var_vals[ji.prg.p.vars[i]])
+      for i in range(len(ji.prg.p.vars)): self.ji_kargs_structs[j].__setattr__(f'v{i}', var_vals[ji.prg.p.vars[i].expr])

    # Build queues.
    self.virt_aql_queues: Dict[Compiled, VirtAQLQueue] = {dev:VirtAQLQueue(dev, 2*len(self.jit_cache)+16) for dev in self.devices}
@@ -106,7 +106,7 @@ class HSAGraph(MultiGraphRunner):
    for sig in self.signals_to_reset: hsa.hsa_signal_silent_store_relaxed(sig, 0)
    hsa.hsa_signal_silent_store_relaxed(self.finish_signal, 0)

-  def __call__(self, input_rawbuffers: List[Buffer], var_vals: Dict[Variable, int], wait=False) -> Optional[float]:
+  def __call__(self, input_rawbuffers: List[Buffer], var_vals: Dict[str, int], wait=False) -> Optional[float]:
    # Wait and restore signals
    hsa.hsa_signal_wait_scacquire(self.finish_signal, hsa.HSA_SIGNAL_CONDITION_LT, 1, (1 << 64) - 1, hsa.HSA_WAIT_STATE_ACTIVE)
    for sig in self.signals_to_reset: hsa.hsa_signal_silent_store_relaxed(sig, 1)
@@ -123,7 +123,7 @@ class HSAGraph(MultiGraphRunner):
    # Update var_vals
    for j in self.jc_idx_with_updatable_var_vals:
      for i,v in enumerate(cast(CompiledRunner, self.jit_cache[j].prg).p.vars):
-        self.ji_kargs_structs[j].__setattr__(f'v{i}', var_vals[v])
+        self.ji_kargs_structs[j].__setattr__(f'v{i}', var_vals[v.expr])

    # Update launch dims
    for j in self.jc_idx_with_updatable_launch_dims:
@@ -29,10 +29,10 @@ def uops_to_rdna(function_name:str, uops:UOpGraph) -> str:
  r: Dict[UOp, str] = {}
  for u in uops:
    if u.uop == UOps.SPECIAL:
-      if u.arg[1].startswith("lidx"):
-        r[u] = f'v{u.arg[0]}'
-      elif u.arg[1].startswith("gidx"):
-        r[u] = f's{2+u.arg[0]}'
+      if u.arg.startswith("lidx"):
+        r[u] = f'v{u.src[0].arg}'
+      elif u.arg.startswith("gidx"):
+        r[u] = f's{2+u.src[0].arg}'
      else:
        raise NotImplementedError
    elif u.uop == UOps.CONST:
@@ -10,7 +10,7 @@ from tinygrad.uop.ops import Ops
 import json
 from collections import OrderedDict

-EXPORT_SUPPORTED_DEVICE = ["WEBGPU", "CPU", "CUDA", "GPU"]
+EXPORT_SUPPORTED_DEVICE = ["WEBGPU", "CPU", "CUDA", "CL"]

 def compile_net(run:TinyJit, special_names:Dict[int,str]) -> Tuple[Dict[str,str],List[Tuple[str,List[str],List[int]]],Dict[str,Tuple[int,DType,int]],Dict[str,Tensor]]:
  functions, bufs, bufs_to_save, statements, bufnum = {}, {}, {}, [], 0
@@ -67,11 +67,12 @@ def export_model_clang(functions:Dict[str,str], statements:Dict[str,Tuple[str,in
  forward_args = ",".join(f"{dtype}{'*' if name not in symbolic_vars.values() else ''} {name}" for name,dtype,_ in (outputs+inputs if wasm else inputs+outputs))

  if not wasm:
+    thread_id = 0 # NOTE: export does not support threading, thread_id is always 0
    for name,cl in bufs_to_save.items():
      weight = ''.join(["\\x%02X"%x for x in bytes(to_mv(cl._buf.va_addr, cl._buf.size))])
      cprog.append(f"unsigned char {name}_data[] = \"{weight}\";")
    cprog += [f"{dtype_map[dtype]} {name}[{len}];" if name not in bufs_to_save else f"{dtype_map[dtype]} *{name} = ({dtype_map[dtype]} *){name}_data;" for name,(len,dtype,_key) in bufs.items() if name not in input_names+output_names]
-    cprog += [f"void net({forward_args}) {{"] + [f"{name}({', '.join(args)});" for (name, args, _global_size, _local_size) in statements] + ["}"]
+    cprog += [f"void net({forward_args}) {{"] + [f"{name}({', '.join(args)}, {thread_id});" for (name, args, _global_size, _local_size) in statements] + ["}"]
    return '\n'.join(headers + cprog)
  else:
    if bufs_to_save:
@@ -239,7 +240,9 @@ export default {model_name};

 def export_model(model, target:str, *inputs, model_name: Optional[str] = "model", stream_weights=False):
  assert Device.DEFAULT in EXPORT_SUPPORTED_DEVICE, f"only {', '.join(EXPORT_SUPPORTED_DEVICE)} are supported"
-  with Context(JIT=2): run,special_names = jit_model(model, *inputs)
+
+  # NOTE: CPU_COUNT=1, since export does not support threading
+  with Context(JIT=2, CPU_COUNT=1): run,special_names = jit_model(model, *inputs)
  functions, statements, bufs, bufs_to_save = compile_net(run, special_names)
  state = get_state_dict(model)
  weight_names = {id(x.uop.base.realized): name for name, x in state.items()}
@@ -5,7 +5,7 @@ from tinygrad.dtype import AddrSpace
 from tinygrad.helpers import getenv, colored, prod, unwrap
 from tinygrad.shape.shapetracker import ShapeTracker, View
 from tinygrad.shape.view import strides_for_shape
-from tinygrad.codegen.opt.kernel import axis_colors
+from tinygrad.codegen.opt.kernel import axis_colors, Opt, OptOps
 from tinygrad.codegen.opt.swizzler import merge_views, view_left

 def to_colored(full_shape, axis_types): return '_'.join([colored(str(s), axis_colors[at]) for s,at in zip(full_shape, axis_types)])
@@ -44,13 +44,28 @@ pm = PatternMatcher([
  (UPat(Ops.VIEW, src=(UPat(Ops.REDUCE_AXIS, src=(UPat.var("src"),), name="r"),), name="view"), swizzle_reduceop),
 ])

+def rangeify_kernel3():
+  a = Tensor.empty(N,N)
+  b = Tensor.empty(N,N)
+  c = a@b
+  #c = c.reshape((32,2,16,4,32,2,16,4)).contiguous()
+  with Context(RANGEIFY=1):
+    sink = c.schedule()[-1].ast
+  #print(sink)
+
+  opts  = [Opt(OptOps.UPCAST, 0, 4), Opt(OptOps.LOCAL, 0, 16), Opt(OptOps.UPCAST, 0, 2)]
+  opts += [Opt(OptOps.UPCAST, 1, 4), Opt(OptOps.LOCAL, 1, 16), Opt(OptOps.UPCAST, 1, 2)]
+  opts += [Opt(OptOps.UNROLL, 0, 8)]
+
+  return sink.replace(arg=KernelInfo(opts_to_apply=tuple(opts)))
+
 def top_spec_kernel3():
  a = Tensor.empty(N,N)
  b = Tensor.empty(N,N)
  c = a@b
  sink = c.schedule()[-1].ast
  L = 16
-  sink = sink.reshape((N//L, L, N//L, L)) #.lift({0:UOp.range(dtypes.int, N//BM, 0), 2:UOp.range(dtypes.int, N//BN, 1)})
+  sink = sink.reshape((N//L, L, N//L, L)) #.lift({0:UOp.range(N//BM, 0), 2:UOp.range(N//BN, 1)})
  sink = graph_rewrite(sink, view_left+pm)
  axis_types = (AxisType.GLOBAL, AxisType.LOCAL, AxisType.GLOBAL, AxisType.LOCAL, AxisType.REDUCE)
  return sink.replace(arg=KernelInfo(name="top_"+to_colored(sink.full_shape, axis_types), axis_types=axis_types))
@@ -171,7 +186,7 @@ def hand_spec_kernel3(kernel4=getenv("K4", 0), kernel5=getenv("K5", 0)):

  c_regs = UOp(Ops.DEFINE_REG, dtypes.float.ptr(TM * nbIterWaveM * TN * nbIterWaveN), arg=2)

-  i = UOp.range(dtypes.int, c_regs.dtype.size, 16)
+  i = UOp.range(c_regs.dtype.size, 16)
  init_store = c_regs[i].store(UOp.const(dtypes.float, 0.0), i)

  if kernel4:
@@ -182,53 +197,53 @@ def hand_spec_kernel3(kernel4=getenv("K4", 0), kernel5=getenv("K5", 0)):
    kId = 0

    # load from globals into locals
-    i = UOp.range(dtypes.int, nbReadsB, 0)
+    i = UOp.range(nbReadsB, 0)
    index_x = BN * blockIdx_x + rBIdx
    index_y = rBIdy + i * strideReadB + kId
    Bs_store = Bs[(index_y % BK) * BN + index_x % BN].store(b[N * index_y + index_x].load(), i)

-    i = UOp.range(dtypes.int, nbReadsA, 1)
+    i = UOp.range(nbReadsA, 1)
    index_x = rAIdx + kId
    index_y = BM * blockIdx_y + rAIdy + i * strideReadA
    As_store = As[(index_x % BK) * BM_As_stride + index_y % BM].store(a[N * index_y + index_x].load(), i)

    # iterate over the middle chunk
-    kId_range = UOp.range(dtypes.int, N//BK-1, 2)
+    kId_range = UOp.range(N//BK-1, 2)
    kId = kId_range*BK

    barrier = UOp.barrier(As_store, Bs_store)

    # load from globals into registers (next round)
-    i = UOp.range(dtypes.int, nbReadsB, 3)
+    i = UOp.range(nbReadsB, 3)
    index_x = BN * blockIdx_x + rBIdx
    index_y = rBIdy + i * strideReadB + kId + BK
    regB_store = regB[i].store(b[N * index_y + index_x].load(), i)

-    i = UOp.range(dtypes.int, nbReadsA, 4)
+    i = UOp.range(nbReadsA, 4)
    index_x = rAIdx + kId + BK
    index_y = BM * blockIdx_y + rAIdy + i * strideReadA
    regA_store = regA[i].store(a[N * index_y + index_x].load(), i)

    def inner_loop(first_range, inp_dep=()):
      # inner unroll
-      k = UOp.range(dtypes.int, BK, first_range+0)
+      k = UOp.range(BK, first_range+0)

      # load from locals into registers
-      iterWave = UOp.range(dtypes.int, nbIterWaveN, first_range+1)
-      i = UOp.range(dtypes.int, TN, first_range+2)
+      iterWave = UOp.range(nbIterWaveN, first_range+1)
+      i = UOp.range(TN, first_range+2)
      index = waveIdx * WN + iterWave * SUBWN + TN * idxInWave + i
      B_row_store = B_row[iterWave*TN + i].store(Bs[k*BN + index].load(*inp_dep), iterWave, i)

-      iterWave = UOp.range(dtypes.int, nbIterWaveM, first_range+3)
-      i = UOp.range(dtypes.int, TM, first_range+4)
+      iterWave = UOp.range(nbIterWaveM, first_range+3)
+      i = UOp.range(TM, first_range+4)
      index = waveIdy * WM + iterWave * SUBWM + TM * idyInWave + i
      A_col_store = A_col[iterWave*TM + i].store(As[k*BM_As_stride + index].load(*inp_dep), iterWave, i)

      # do the GEMM math
-      iterWaveM = UOp.range(dtypes.int, nbIterWaveM, first_range+5)
-      yt = UOp.range(dtypes.int, TM, first_range+6)
-      iterWaveN = UOp.range(dtypes.int, nbIterWaveN, first_range+7)
-      xt = UOp.range(dtypes.int, TN, first_range+8)
+      iterWaveM = UOp.range(nbIterWaveM, first_range+5)
+      yt = UOp.range(TM, first_range+6)
+      iterWaveN = UOp.range(nbIterWaveN, first_range+7)
+      xt = UOp.range(TN, first_range+8)
      x = iterWaveN * TN + xt
      y = iterWaveM * TM + yt
      c_regs_idx = c_regs[y * TN * nbIterWaveN + x]
@@ -241,12 +256,12 @@ def hand_spec_kernel3(kernel4=getenv("K4", 0), kernel5=getenv("K5", 0)):
    sink = inner_loop(5, (barrier, regB_store, regA_store)).barrier()

    # load from registers into locals
-    i = UOp.range(dtypes.int, nbReadsB, 14)
+    i = UOp.range(nbReadsB, 14)
    index_x = BN * blockIdx_x + rBIdx
    index_y = rBIdy + i * strideReadB + kId + BK
    Bs_store = Bs[(index_y % BK) * BN + index_x % BN].store(regB[i].load(sink), i, kId_range)

-    i = UOp.range(dtypes.int, nbReadsA, 15)
+    i = UOp.range(nbReadsA, 15)
    index_x = rAIdx + kId + BK
    index_y = BM * blockIdx_y + rAIdy + i * strideReadA
    As_store = As[(index_x % BK) * BM_As_stride + index_y % BM].store(regA[i].load(sink), i, kId_range)
@@ -254,40 +269,40 @@ def hand_spec_kernel3(kernel4=getenv("K4", 0), kernel5=getenv("K5", 0)):
    # final iteration without the copy
    sink = inner_loop(16, (UOp.barrier(Bs_store, As_store),))
  else:
-    kId_range = UOp.range(dtypes.int, N//BK, 0)
+    kId_range = UOp.range(N//BK, 0)
    kId = kId_range*BK

    # load from globals into locals
-    i = UOp.range(dtypes.int, nbReadsB, 1)
+    i = UOp.range(nbReadsB, 1)
    index_x = BN * blockIdx_x + rBIdx
    index_y = rBIdy + i * strideReadB + kId
    Bs_store = Bs[(index_y % BK) * BN + index_x % BN].store(b[N * index_y + index_x].load(), i)

-    i = UOp.range(dtypes.int, nbReadsA, 2)
+    i = UOp.range(nbReadsA, 2)
    index_x = rAIdx + kId
    index_y = BM * blockIdx_y + rAIdy + i * strideReadA
    As_store = As[(index_x % BK) * BM_As_stride + index_y % BM].store(a[N * index_y + index_x].load(), i)

    barrier = UOp.barrier(As_store, Bs_store)

-    k = UOp.range(dtypes.int, BK, 3)
+    k = UOp.range(BK, 3)

    # load from locals into registers
-    iterWave = UOp.range(dtypes.int, nbIterWaveN, 4)
-    i = UOp.range(dtypes.int, TN, 5)
+    iterWave = UOp.range(nbIterWaveN, 4)
+    i = UOp.range(TN, 5)
    index = waveIdx * WN + iterWave * SUBWN + TN * idxInWave + i
    B_row_store = B_row[iterWave*TN + i].store(Bs[k*BN + index].load(barrier), iterWave, i)

-    iterWave = UOp.range(dtypes.int, nbIterWaveM, 6)
-    i = UOp.range(dtypes.int, TM, 7)
+    iterWave = UOp.range(nbIterWaveM, 6)
+    i = UOp.range(TM, 7)
    index = waveIdy * WM + iterWave * SUBWM + TM * idyInWave + i
    A_col_store = A_col[iterWave*TM + i].store(As[k*BM_As_stride + index].load(barrier), iterWave, i)

    # do the GEMM math
-    iterWaveM = UOp.range(dtypes.int, nbIterWaveM, 8)
-    yt = UOp.range(dtypes.int, TM, 9)
-    iterWaveN = UOp.range(dtypes.int, nbIterWaveN, 10)
-    xt = UOp.range(dtypes.int, TN, 12)
+    iterWaveM = UOp.range(nbIterWaveM, 8)
+    yt = UOp.range(TM, 9)
+    iterWaveN = UOp.range(nbIterWaveN, 10)
+    xt = UOp.range(TN, 12)
    x = iterWaveN * TN + xt
    y = iterWaveM * TM + yt
    c_regs_idx = c_regs[y * TN * nbIterWaveN + x]
@@ -295,10 +310,10 @@ def hand_spec_kernel3(kernel4=getenv("K4", 0), kernel5=getenv("K5", 0)):
                            iterWaveM, iterWaveN, yt, xt, k, kId_range)

  # store c_regs into c
-  iterWaveM = UOp.range(dtypes.int, nbIterWaveM, 1000)
-  yt = UOp.range(dtypes.int, TM, 1001)
-  iterWaveN = UOp.range(dtypes.int, nbIterWaveN, 1002)
-  xt = UOp.range(dtypes.int, TN, 1003)
+  iterWaveM = UOp.range(nbIterWaveM, 1000)
+  yt = UOp.range(TM, 1001)
+  iterWaveN = UOp.range(nbIterWaveN, 1002)
+  xt = UOp.range(TN, 1003)
  xOut = blockIdx_x * BN + waveIdx * WN + iterWaveN * SUBWN + TN * idxInWave
  yOut = blockIdx_y * BM + waveIdy * WM + iterWaveM * SUBWM + TM * idyInWave
  indexC = N * (yOut + yt) + xOut + xt
@@ -309,10 +324,15 @@ def hand_spec_kernel3(kernel4=getenv("K4", 0), kernel5=getenv("K5", 0)):

 if __name__ == "__main__":
  HL = getenv("HL")
-  if HL == 2: hprg = top_spec_kernel3()
+  if HL == 3: hprg = rangeify_kernel3()
+  elif HL == 2: hprg = top_spec_kernel3()
  elif HL == 1: hprg = hl_spec_kernel3()
  else: hprg = hand_spec_kernel3()
-  prg = get_program(hprg, Device.default.renderer)
+  if HL == 3:
+    with Context(RANGEIFY=1, BLOCK_REORDER=0):
+      prg = get_program(hprg, Device.default.renderer)
+  else:
+    prg = get_program(hprg, Device.default.renderer)
  print(prg.src)
  if getenv("SRC"): exit(0)
  hrunner = CompiledRunner(prg)
@@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 import numpy as np
-from tinygrad.runtime.ops_gpu import CLProgram, CLCompiler
+from tinygrad.runtime.ops_cl import CLProgram, CLCompiler
 from tinygrad import Device, dtypes
 from tinygrad.device import Buffer
 from hexdump import hexdump
@@ -11,7 +11,7 @@ from hexdump import hexdump
 # https://registry.khronos.org/OpenCL/extensions/intel/cl_intel_subgroup_split_matrix_multiply_accumulate.html
 # https://hc34.hotchips.org/assets/program/conference/day1/GPU%20HPC/Intel_s%20Ponte%20Vecchio%20GPU%20-%20Architecture%20Systems%20and%20Software%20FINAL.pdf

-device = Device["GPU"]
+device = Device["CL"]

 # NOTE: only the subgroup type 8 ones work
 prog = CLProgram(device, "test", CLCompiler(device, "test").compile(f"""
@@ -26,9 +26,9 @@ __kernel void test(__global float* data0, const __global int* data1, const __glo
 """))
 #with open("/tmp/test.elf", "wb") as f: f.write(prog.lib)

-a = Buffer("GPU", 8, dtypes.float32).allocate()
-b = Buffer("GPU", 0x10, dtypes.float16).allocate()
-c = Buffer("GPU", 8*0x10, dtypes.float16).allocate()
+a = Buffer("CL", 8, dtypes.float32).allocate()
+b = Buffer("CL", 0x10, dtypes.float16).allocate()
+c = Buffer("CL", 8*0x10, dtypes.float16).allocate()

 row = np.array([1,2,3,4,5,6,7,8,1,2,3,4,5,6,7,8], np.float16)
 mat = np.random.random((8, 0x10)).astype(np.float16)
@@ -56,7 +56,7 @@ def randoms():
 def ast_to_cuda_prog(compiler, ast, opts):
  k = Kernel(ast)
  k.apply_opts(opts)
-  p = get_program(k.get_optimized_ast(), k.opts)
+  p = get_program(k.ast, k.opts, k.applied_opts)
  return CUDAProgram(device, p.function_name, compiler.compile(p.src))

 if __name__ == "__main__":
@@ -75,7 +75,7 @@ if __name__ == "__main__":

    if GEMM_VARIATION == "max" and (M%64)==0 and (N%128)==0 and (K%64)==0 and DTYPE_IN == dtypes.half and DTYPE_OUT == dtypes.float and DTYPE_ACC == dtypes.float:
      print("Using CUDA and triton-generated kernel")
-      # See nv_triton_gemm.annotated.ptx for PTX code which was generated from `PYTHONPATH=. DEBUG=6 CUDA=1 PTX=1 python3 extra/gemm/triton_nv_matmul.py`
+      # See nv_triton_gemm.annotated.ptx for PTX code which was generated from `PYTHONPATH=. DEBUG=6 CUDA=1 CUDA_PTX=1 python3 extra/gemm/triton_nv_matmul.py`
      # this kernel with M=N=K=4096 does 162TFLOPS, vs torch at 144TFLOPS and BEAM=8 tinygrad at 138TFLOPS.  theo max is 165TFLOPS.

      # WMMA element size is (M, N, K) = (16, 8, 16)
@@ -2,7 +2,7 @@ import numpy as np
 from tinygrad import dtypes, Tensor
 from tinygrad.helpers import getenv, get_single_element
 from tinygrad.dtype import _to_np_dtype
-from tinygrad.codegen.opt.kernel import OptOps
+from tinygrad.codegen.opt import OptOps
 from tinygrad.engine.realize import lower_schedule

 dtype_in = dtypes.half if getenv("HALF") else dtypes.bfloat16 if getenv("BFLOAT16") else dtypes.float
@@ -29,7 +29,7 @@ if __name__ == "__main__":
      Opt(op=OptOps.LOCAL, axis=0, amt=2),
    ]
  k.apply_opts(opts)
-  prg = get_program(k.get_optimized_ast(), k.opts)
+  prg = get_program(k.ast, k.opts, k.applied_opts)
  new_src = prg.src
  # can mod source here
  prg = replace(prg, src=new_src)
@@ -43,7 +43,7 @@ def matmul_kernel(c_ptr, a_ptr, b_ptr, BLOCK_SIZE_M: tl.constexpr, BLOCK_SIZE_N:
  c_ptrs = c_ptr + stride_cm * offs_cm[:, None] + stride_cn * offs_cn[None, :]
  tl.store(c_ptrs, c)

-# CUDA=1 PTX=1 python3 extra/gemm/triton_nv_matmul.py
+# CUDA=1 CUDA_PTX=1 python3 extra/gemm/triton_nv_matmul.py
 if __name__ == "__main__":
  BLOCK_SIZE_M, BLOCK_SIZE_N, BLOCK_SIZE_K = 64, 128, 64
  M, N, K = 4096, 4096, 4096
@@ -7,7 +7,6 @@ bert_train_params = {
  "GPUS": 6,
  "BS": 96,
  "EVAL_BS": 96,
-  "FUSE_ARANGE": 1,
  "BASEDIR": "/raid/datasets/wiki",
 }

@@ -50,7 +50,7 @@ def ioctls_from_header():
  hdr = (pathlib.Path(__file__).parent / "kfd_ioctl.h").read_text().replace("\\\n", "")
  pattern = r'#define\s+(AMDKFD_IOC_[A-Z0-9_]+)\s+AMDKFD_IOW?R?\((0x[0-9a-fA-F]+),\s+struct\s([A-Za-z0-9_]+)\)'
  matches = re.findall(pattern, hdr, re.MULTILINE)
-  return {int(nr, 0x10):(name, getattr(kfd_ioctl, "struct_"+sname)) for name, nr, sname in matches}
+  return {int(nr, 0x10):(name, getattr(kfd_ioctl, "struct_"+sname, None)) for name, nr, sname in matches}
 nrs = ioctls_from_header()

@ctypes.CFUNCTYPE(ctypes.c_int, ctypes.c_int, ctypes.c_ulong, ctypes.c_void_p)
@@ -0,0 +1,61 @@
+# HuggingFace ONNX
+
+Tool for discovering, downloading, and validating ONNX models from HuggingFace.
+
+## Extra Dependencies
+
+```bash
+pip install huggingface_hub pyyaml requests onnx onnxruntime numpy
+```
+
+## Huggingface Manager (discovering and downloading)
+
+The `huggingface_manager.py` script discovers top ONNX models from HuggingFace, collects metadata, and optionally downloads them.
+
+```bash
+# Download top 50 models sorted by downloads
+python huggingface_manager.py --limit 50 --download
+
+# Just collect metadata (no download)
+python huggingface_manager.py --limit 100
+
+# Sort by likes instead of downloads
+python huggingface_manager.py --limit 20 --sort likes --download
+
+# Custom output file
+python huggingface_manager.py --limit 10 --output my_models.yaml
+```
+
+### Output Format
+
+The tool generates a YAML file with the following structure:
+
+```yaml
+repositories:
+  "model-name":
+    url: "https://huggingface.co/model-name"
+    download_path: "/path/to/models/..."  # when --download used
+    files:
+      - file: "model.onnx"
+        size: "90.91MB"
+total_size: "2.45GB"
+created_at: "2024-01-15T10:30:00Z"
+```
+
+## Run Models (validation)
+
+The `run_models.py` script validates ONNX models against ONNX Runtime for correctness.
+
+```bash
+# Validate models from a YAML configuration file
+python run_models.py --validate huggingface_repos.yaml
+
+# Debug specific repository (downloads and validates all ONNX models)
+python run_models.py --debug sentence-transformers/all-MiniLM-L6-v2
+
+# Debug specific model file
+python run_models.py --debug sentence-transformers/all-MiniLM-L6-v2/onnx/model.onnx
+
+# Debug with model truncation for debugging and validating intermediate results
+DEBUGONNX=1 python run_models.py --debug sentence-transformers/all-MiniLM-L6-v2/onnx/model.onnx --truncate 10
+```
@@ -1,85 +0,0 @@
-import yaml, time, requests, argparse
-from pathlib import Path
-from huggingface_hub import list_models, HfApi
-from tinygrad.helpers import tqdm
-
-HUGGINGFACE_URL = "https://huggingface.co"
-SKIPPED_FILES = [
-  "fp16", "int8", "uint8", "quantized",      # numerical accuracy issues
-  "avx2", "arm64", "avx512", "avx512_vnni",  # numerical accuracy issues
-  "q4", "q4f16", "bnb4",                     # unimplemented quantization
-  "model_O4",                                # requires non cpu ort runner and MemcpyFromHost op
-  "merged",                                  # TODO implement attribute with graph type and Loop op
-]
-SKIPPED_REPO_PATHS = [
-  # Invalid model-index
-  "AdamCodd/vit-base-nsfw-detector",
-  # TODO: implement attribute with graph type and Loop op
-  "minishlab/potion-base-8M", "minishlab/M2V_base_output", "minishlab/potion-retrieval-32M",
-  # TODO: implement SimplifiedLayerNormalization, SkipSimplifiedLayerNormalization, GroupQueryAttention
-  "HuggingFaceTB/SmolLM2-360M-Instruct",
-  # TODO: implement SimplifiedLayerNormalization, SkipSimplifiedLayerNormalization, RotaryEmbedding, MultiHeadAttention
-  "HuggingFaceTB/SmolLM2-1.7B-Instruct",
-  # TODO: implmement RandomNormalLike
-  "stabilityai/stable-diffusion-xl-base-1.0", "stabilityai/sdxl-turbo", 'SimianLuo/LCM_Dreamshaper_v7',
-  # TODO: implement NonZero
-  "mangoapps/fb_zeroshot_mnli_onnx",
-  # TODO huge Concat in here with 1024 (1, 3, 32, 32) Tensors, and maybe a MOD bug with const folding
-  "briaai/RMBG-2.0",
-]
-
-def get_top_repos(n: int, sort: str) -> list[str]: # list["FacebookAI/xlm-roberta-large", ...]
-  print(f"** Getting top {n} models sorted by {sort} **")
-  repos = []
-  i = 0
-  for model in list_models(filter="onnx", sort=sort):
-    if model.id in SKIPPED_REPO_PATHS: continue
-    print(f"{i+1}/{n}: {model.id} ({getattr(model, sort)})")
-    repos.append(model.id)
-    i += 1
-    if i == n: break
-  return repos
-
-def get_metadata(repos:list[str]) -> dict:
-  api = HfApi()
-  repos_metadata = {"repositories": {}}
-  total_size = 0
-
-  # TODO: speed head requests up with async?
-  for repo in tqdm(repos, desc="Getting metadata"):
-    files_metadata = []
-    model_info = api.model_info(repo)
-
-    for file in model_info.siblings:
-      filename = file.rfilename
-      if not (filename.endswith('.onnx') or filename.endswith('.onnx_data')): continue
-      if any(skip_str in filename for skip_str in SKIPPED_FILES): continue
-      head = requests.head(f"{HUGGINGFACE_URL}/{repo}/resolve/main/{filename}", allow_redirects=True)
-      file_size = file.size or int(head.headers.get('Content-Length', 0))
-      files_metadata.append({"file": filename, "size": f"{file_size/1e6:.2f}MB"})
-      total_size += file_size
-
-    repos_metadata["repositories"][repo] = {
-      "url": f"{HUGGINGFACE_URL}/{repo}",
-      "download_path": None,
-      "files": files_metadata,
-    }
-  repos_metadata['total_size'] = f"{total_size/1e9:.2f}GB"
-  repos_metadata['created_at'] = time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime())
-  return repos_metadata
-
-if __name__ == "__main__":
-  sort = "downloads" # recent 30 days downloads
-  huggingface_onnx_dir = Path(__file__).parent
-
-  parser = argparse.ArgumentParser(description="Produces a YAML file with metadata of top huggingface onnx models")
-  parser.add_argument("--limit", type=int, required=True, help="Number of top repositories to process (e.g., 100)")
-  parser.add_argument("--output", type=str, default="huggingface_repos.yaml", help="Output YAML file name to save the report")
-  args = parser.parse_args()
-
-  top_repos = get_top_repos(args.limit, sort)
-  metadata = get_metadata(top_repos)
-  yaml_path = huggingface_onnx_dir / args.output
-  with open(yaml_path, 'w') as f:
-    yaml.dump(metadata, f, sort_keys=False)
-    print(f"YAML saved to: {str(yaml_path)}")
@@ -1,29 +0,0 @@
-import yaml, argparse
-from pathlib import Path
-from huggingface_hub import snapshot_download
-
-def download_models(yaml_file: str, download_dir: str) -> None:
-  with open(yaml_file, 'r') as f: metadata = yaml.safe_load(f)
-  n = len(metadata["repositories"])
-
-  for i, (model_id, model_data) in enumerate(metadata["repositories"].items()):
-    print(f"Downloading {i+1}/{n}: {model_id}...")
-    allow_patterns = [file_info["file"] for file_info in model_data["files"]]
-    root_path = Path(snapshot_download(repo_id=model_id, allow_patterns=allow_patterns, cache_dir=download_dir))
-    # download configs too (the sizes are small)
-    snapshot_download(repo_id=model_id, allow_patterns=["*config.json"], cache_dir=download_dir)
-    print(f"Downloaded model files to: {root_path}")
-    model_data["download_path"] = str(root_path)
-
-  # Save the updated metadata back to the YAML file
-  with open(yaml_file, 'w') as f: yaml.dump(metadata, f, sort_keys=False)
-  print("Download completed according to YAML file.")
-
-if __name__ == "__main__":
-  parser = argparse.ArgumentParser(description="Download models from Huggingface Hub based on a YAML configuration file.")
-  parser.add_argument("input", type=str, help="Path to the input YAML configuration file containing model information.")
-  args = parser.parse_args()
-
-  models_folder = Path(__file__).parent / "models"
-  models_folder.mkdir(parents=True, exist_ok=True)
-  download_models(args.input, str(models_folder))
@@ -0,0 +1,230 @@
+import yaml
+import time
+import requests
+import argparse
+from pathlib import Path
+from huggingface_hub import list_models, HfApi, snapshot_download
+from tinygrad.helpers import _ensure_downloads_dir
+DOWNLOADS_DIR = _ensure_downloads_dir() / "models"
+from tinygrad.helpers import tqdm
+
+def snapshot_download_with_retry(*, repo_id: str, allow_patterns: list[str]|tuple[str, ...]|None=None, cache_dir: str|Path|None=None,
+                                 tries: int=2, **kwargs) -> Path:
+  for attempt in range(tries):
+    try:
+      return Path(snapshot_download(
+        repo_id=repo_id,
+        allow_patterns=allow_patterns,
+        cache_dir=str(cache_dir) if cache_dir is not None else None,
+        **kwargs
+      ))
+    except Exception as e:
+      if attempt == tries-1: raise
+      time.sleep(1)
+
+# Constants for filtering models
+HUGGINGFACE_URL = "https://huggingface.co"
+SKIPPED_FILES = [
+  "fp16", "int8", "uint8", "quantized",      # numerical accuracy issues
+  "avx2", "arm64", "avx512", "avx512_vnni",  # numerical accuracy issues
+  "q4", "q4f16", "bnb4",                     # unimplemented quantization
+  "model_O4",                                # requires non cpu ort runner and MemcpyFromHost op
+  "merged",                                  # TODO implement attribute with graph type and Loop op
+]
+
+SKIPPED_REPO_PATHS = [
+  # Invalid model-index
+  "AdamCodd/vit-base-nsfw-detector",
+  # TODO: implement attribute with graph type and Loop op
+  "minishlab/potion-base-8M", "minishlab/M2V_base_output", "minishlab/potion-retrieval-32M",
+  # TODO: implement SimplifiedLayerNormalization, SkipSimplifiedLayerNormalization, GroupQueryAttention
+  "HuggingFaceTB/SmolLM2-360M-Instruct",
+  # TODO: implement SimplifiedLayerNormalization, SkipSimplifiedLayerNormalization, RotaryEmbedding, MultiHeadAttention
+  "HuggingFaceTB/SmolLM2-1.7B-Instruct",
+  # TODO: implement RandomNormalLike
+  "stabilityai/stable-diffusion-xl-base-1.0", "stabilityai/sdxl-turbo", 'SimianLuo/LCM_Dreamshaper_v7',
+  # TODO: implement NonZero
+  "mangoapps/fb_zeroshot_mnli_onnx",
+  # TODO huge Concat in here with 1024 (1, 3, 32, 32) Tensors, and maybe a MOD bug with const folding
+  "briaai/RMBG-2.0",
+]
+
+
+class HuggingFaceONNXManager:
+  def __init__(self):
+    self.base_dir = Path(__file__).parent
+    self.models_dir = DOWNLOADS_DIR
+    self.api = HfApi()
+
+  def discover_models(self, limit: int, sort: str = "downloads") -> list[str]:
+    print(f"Discovering top {limit} ONNX models sorted by {sort}...")
+    repos = []
+    i = 0
+
+    for model in list_models(filter="onnx", sort=sort):
+      if model.id in SKIPPED_REPO_PATHS:
+        continue
+
+      print(f"  {i+1}/{limit}: {model.id} ({getattr(model, sort)})")
+      repos.append(model.id)
+      i += 1
+      if i == limit:
+        break
+
+    print(f"Found {len(repos)} suitable ONNX models")
+    return repos
+
+  def collect_metadata(self, repos: list[str]) -> dict:
+    print(f"Collecting metadata for {len(repos)} repositories...")
+    metadata = {"repositories": {}}
+    total_size = 0
+
+    for repo in tqdm(repos, desc="Collecting metadata"):
+      try:
+        files_metadata = []
+        model_info = self.api.model_info(repo)
+
+        for file in model_info.siblings:
+          filename = file.rfilename
+          if not (filename.endswith('.onnx') or filename.endswith('.onnx_data')):
+            continue
+          if any(skip_str in filename for skip_str in SKIPPED_FILES):
+            continue
+
+          # Get file size from API or HEAD request
+          try:
+            head = requests.head(
+              f"{HUGGINGFACE_URL}/{repo}/resolve/main/{filename}",
+              allow_redirects=True,
+              timeout=10
+            )
+            file_size = file.size or int(head.headers.get('Content-Length', 0))
+          except requests.RequestException:
+            file_size = file.size or 0
+
+          files_metadata.append({
+            "file": filename,
+            "size": f"{file_size/1e6:.2f}MB"
+          })
+          total_size += file_size
+
+        if files_metadata:  # Only add repos with valid ONNX files
+          metadata["repositories"][repo] = {
+            "url": f"{HUGGINGFACE_URL}/{repo}",
+            "download_path": None,
+            "files": files_metadata,
+          }
+
+      except Exception as e:
+        print(f"WARNING: Failed to collect metadata for {repo}: {e}")
+        continue
+
+    metadata['total_size'] = f"{total_size/1e9:.2f}GB"
+    metadata['created_at'] = time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime())
+
+    print(f"Collected metadata for {len(metadata['repositories'])} repositories")
+    print(f"Total estimated download size: {metadata['total_size']}")
+
+    return metadata
+
+  def download_models(self, metadata: dict) -> dict:
+    self.models_dir.mkdir(parents=True, exist_ok=True)
+
+    repos = metadata["repositories"]
+    n = len(repos)
+
+    print(f"Downloading {n} repositories to {self.models_dir}...")
+
+    for i, (model_id, model_data) in enumerate(repos.items()):
+      print(f"  Downloading {i+1}/{n}: {model_id}...")
+
+      try:
+        # Download ONNX model files
+        allow_patterns = [file_info["file"] for file_info in model_data["files"]]
+        root_path = snapshot_download_with_retry(
+          repo_id=model_id,
+          allow_patterns=allow_patterns,
+          cache_dir=str(self.models_dir)
+        )
+
+        # Download config files (usually small)
+        snapshot_download_with_retry(
+          repo_id=model_id,
+          allow_patterns=["*config.json"],
+          cache_dir=str(self.models_dir)
+        )
+
+        model_data["download_path"] = str(root_path)
+        print(f"    Downloaded to: {root_path}")
+
+      except Exception as e:
+        print(f"    ERROR: Failed to download {model_id}: {e}")
+        model_data["download_path"] = None
+        continue
+
+    successful_downloads = sum(1 for repo in repos.values() if repo["download_path"] is not None)
+    print(f"Successfully downloaded {successful_downloads}/{n} repositories")
+    print(f"All models saved to: {self.models_dir}")
+
+    return metadata
+
+  def save_metadata(self, metadata: dict, output_file: str):
+    yaml_path = self.base_dir / output_file
+    with open(yaml_path, 'w') as f:
+      yaml.dump(metadata, f, sort_keys=False)
+    print(f"Metadata saved to: {yaml_path}")
+
+  def discover_and_download(self, limit: int, output_file: str = "huggingface_repos.yaml",
+                          sort: str = "downloads", download: bool = True):
+    print(f"Starting HuggingFace ONNX workflow...")
+    print(f"   Limit: {limit} models")
+    print(f"   Sort by: {sort}")
+    print(f"   Download: {'Yes' if download else 'No'}")
+    print(f"   Output: {output_file}")
+    print("-" * 50)
+
+    repos = self.discover_models(limit, sort)
+
+    metadata = self.collect_metadata(repos)
+
+    if download:
+      metadata = self.download_models(metadata)
+
+    self.save_metadata(metadata, output_file)
+
+    print("-" * 50)
+    print("Workflow completed successfully!")
+    if download:
+      successful = sum(1 for repo in metadata["repositories"].values()
+                     if repo["download_path"] is not None)
+      print(f"{successful}/{len(metadata['repositories'])} models downloaded")
+
+    return metadata
+
+
+if __name__ == "__main__":
+  parser = argparse.ArgumentParser(
+    description="HuggingFace ONNX Model Manager - Discover, collect metadata, and download ONNX models",
+  )
+
+  parser.add_argument("--limit", type=int, help="Number of top repositories to process")
+  parser.add_argument("--output", type=str, default="huggingface_repos.yaml",
+                     help="Output YAML file name (default: huggingface_repos.yaml)")
+  parser.add_argument("--sort", type=str, default="downloads",
+                     choices=["downloads", "likes", "created", "modified"],
+                     help="Sort criteria for model discovery (default: downloads)")
+
+  parser.add_argument("--download", action="store_true", default=False,
+                     help="Download models after collecting metadata")
+
+  args = parser.parse_args()
+
+  if not args.limit: parser.error("--limit is required")
+
+  manager = HuggingFaceONNXManager()
+  manager.discover_and_download(
+    limit=args.limit,
+    output_file=args.output,
+    sort=args.sort,
+    download=args.download
+  )
@@ -1,10 +1,11 @@
-import onnx, yaml, tempfile, time, collections, pprint, argparse, json
+import onnx, yaml, tempfile, time, argparse, json
 from pathlib import Path
-from tinygrad.frontend.onnx import OnnxRunner
-from extra.onnx import get_onnx_ops
+from typing import Any
+from tinygrad.nn.onnx import OnnxRunner
 from extra.onnx_helpers import validate, get_example_inputs
+from extra.huggingface_onnx.huggingface_manager import DOWNLOADS_DIR, snapshot_download_with_retry

-def get_config(root_path: Path):
+def get_config(root_path: Path) -> dict[str, Any]:
  ret = {}
  for path in root_path.rglob("*config.json"):
    config = json.load(path.open())
@@ -12,19 +13,19 @@ def get_config(root_path: Path):
      ret.update(config)
  return ret

-def run_huggingface_validate(onnx_model_path, config, rtol, atol):
-  onnx_runner = OnnxRunner(onnx_model_path)
-  inputs = get_example_inputs(onnx_runner.graph_inputs, config)
-  validate(onnx_model_path, inputs, rtol=rtol, atol=atol)
-
-def get_tolerances(file_name): # -> rtol, atol
+def get_tolerances(file_name: str) -> tuple[float, float]:
  # TODO very high rtol atol
  if "fp16" in file_name: return 9e-2, 9e-2
  if any(q in file_name for q in ["int8", "uint8", "quantized"]): return 4, 4
  return 4e-3, 3e-2

+def run_huggingface_validate(onnx_model_path: str | Path, config: dict[str, Any], rtol: float, atol: float):
+  onnx_runner = OnnxRunner(onnx_model_path)
+  inputs = get_example_inputs(onnx_runner.graph_inputs, config)
+  validate(onnx_model_path, inputs, rtol=rtol, atol=atol)
+
 def validate_repos(models:dict[str, tuple[Path, Path]]):
-  print(f"** Validating {len(model_paths)} models **")
+  print(f"** Validating {len(models)} models **")
  for model_id, (root_path, relative_path) in models.items():
    print(f"validating model {model_id}")
    model_path = root_path / relative_path
@@ -36,25 +37,6 @@ def validate_repos(models:dict[str, tuple[Path, Path]]):
    et = time.time() - st
    print(f"passed, took {et:.2f}s")

-def retrieve_op_stats(models:dict[str, tuple[Path, Path]]) -> dict:
-  ret = {}
-  op_counter = collections.Counter()
-  unsupported_ops = collections.defaultdict(set)
-  supported_ops = get_onnx_ops()
-  print(f"** Retrieving stats from {len(model_paths)} models **")
-  for model_id, (root_path, relative_path) in models.items():
-    print(f"examining {model_id}")
-    model_path = root_path / relative_path
-    onnx_runner = OnnxRunner(model_path)
-    for node in onnx_runner.graph_nodes:
-      op_counter[node.op] += 1
-      if node.op not in supported_ops:
-        unsupported_ops[node.op].add(model_id)
-    del onnx_runner
-  ret["unsupported_ops"] = {k:list(v) for k, v in unsupported_ops.items()}
-  ret["op_counter"] = op_counter.most_common()
-  return ret
-
 def debug_run(model_path, truncate, config, rtol, atol):
  if truncate != -1:
    model = onnx.load(model_path)
@@ -71,12 +53,9 @@ def debug_run(model_path, truncate, config, rtol, atol):
    run_huggingface_validate(model_path, config, rtol, atol)

 if __name__ == "__main__":
-  parser = argparse.ArgumentParser(description="Huggingface ONNX Model Validator and Ops Checker")
-  parser.add_argument("input", type=str, help="Path to the input YAML configuration file containing model information.")
-  parser.add_argument("--check_ops", action="store_true", default=False,
-                      help="Check support for ONNX operations in models from the YAML file")
-  parser.add_argument("--validate", action="store_true", default=False,
-                      help="Validate correctness of models from the YAML file")
+  parser = argparse.ArgumentParser(description="Huggingface ONNX Model Validator")
+  parser.add_argument("--validate", type=str, default="",
+                      help="Validate correctness of models from the specified YAML configuration file")
  parser.add_argument("--debug", type=str, default="",
                      help="""Validates without explicitly needing a YAML or models pre-installed.
                      provide repo id (e.g. "minishlab/potion-base-8M") to validate all onnx models inside the repo
@@ -85,13 +64,13 @@ if __name__ == "__main__":
  parser.add_argument("--truncate", type=int, default=-1, help="Truncate the ONNX model so intermediate results can be validated")
  args = parser.parse_args()

-  if not (args.check_ops or args.validate or args.debug):
-    parser.error("Please provide either --validate, --check_ops, or --debug.")
+  if not (args.validate or args.debug):
+    parser.error("Please provide either --validate <yaml_file> or --debug <repo_id>.")
  if args.truncate != -1 and not args.debug:
    parser.error("--truncate and --debug should be used together for debugging")

-  if args.check_ops or args.validate:
-    with open(args.input, 'r') as f:
+  if args.validate:
+    with open(args.validate, 'r') as f:
      data = yaml.safe_load(f)
      assert all(repo["download_path"] is not None for repo in data["repositories"].values()), "please run `download_models.py` for this yaml"
      model_paths = {
@@ -101,22 +80,16 @@ if __name__ == "__main__":
        if model["file"].endswith(".onnx")
      }

-    if args.check_ops:
-      pprint.pprint(retrieve_op_stats(model_paths))
-
-    if args.validate:
-      validate_repos(model_paths)
+    validate_repos(model_paths)

  if args.debug:
-    from huggingface_hub import snapshot_download
-    download_dir = Path(__file__).parent / "models"
    path:list[str] = args.debug.split("/")
    if len(path) == 2:
      # repo id
      # validates all onnx models inside repo
      repo_id = "/".join(path)
-      root_path = Path(snapshot_download(repo_id=repo_id, allow_patterns=["*.onnx", "*.onnx_data"], cache_dir=download_dir))
-      snapshot_download(repo_id=repo_id, allow_patterns=["*config.json"], cache_dir=download_dir)
+      root_path = snapshot_download_with_retry(repo_id=repo_id, allow_patterns=["*.onnx", "*.onnx_data"], cache_dir=DOWNLOADS_DIR)
+      snapshot_download_with_retry(repo_id=repo_id, allow_patterns=["*config.json"], cache_dir=DOWNLOADS_DIR)
      config = get_config(root_path)
      for onnx_model in root_path.rglob("*.onnx"):
        rtol, atol = get_tolerances(onnx_model.name)
@@ -128,8 +101,8 @@ if __name__ == "__main__":
      onnx_model = path[-1]
      assert path[-1].endswith(".onnx")
      repo_id, relative_path = "/".join(path[:2]), "/".join(path[2:])
-      root_path = Path(snapshot_download(repo_id=repo_id, allow_patterns=[relative_path], cache_dir=download_dir))
-      snapshot_download(repo_id=repo_id, allow_patterns=["*config.json"], cache_dir=download_dir)
+      root_path = snapshot_download_with_retry(repo_id=repo_id, allow_patterns=[relative_path], cache_dir=DOWNLOADS_DIR)
+      snapshot_download_with_retry(repo_id=repo_id, allow_patterns=["*config.json"], cache_dir=DOWNLOADS_DIR)
      config = get_config(root_path)
      rtol, atol = get_tolerances(onnx_model)
      print(f"validating {relative_path} with truncate={args.truncate}, {rtol=}, {atol=}")
@@ -88,7 +88,7 @@ def mcts_search(lin:Kernel, rawbufs:List[Buffer], amt:int) -> Kernel:
    return ret

  rawbufs = _ensure_buffer_alloc(rawbufs)
-  var_vals = {k:(k.vmax+k.vmin)//2 for k in lin.ast.variables()}
+  var_vals = {k.expr:(k.vmax+k.vmin)//2 for k in lin.ast.variables()}
  dev = Device[lin.opts.device]
  root = MCTSNode(lin)

@@ -9,6 +9,9 @@ from PIL import Image
 import numpy as np
 import re, gzip

+# Allow for monkeypatching for mlperf.
+gelu = Tensor.gelu
+
@lru_cache()
 def default_bpe():
  # Clip tokenizer, taken from https://github.com/openai/CLIP/blob/main/clip/simple_tokenizer.py (MIT license)
@@ -53,8 +56,8 @@ class Tokenizer:
    cs = [chr(n) for n in cs]
    return dict(zip(bs, cs))
  class ClipTokenizer:
-    def __init__(self):
-      self.byte_encoder = Tokenizer.bytes_to_unicode()
+    def __init__(self, version=None):
+      self.byte_encoder, self.version = Tokenizer.bytes_to_unicode(), version
      merges = gzip.open(default_bpe()).read().decode("utf-8").split('\n')
      merges = merges[1:49152-256-2+1]
      merges = [tuple(merge.split()) for merge in merges]
@@ -62,11 +65,17 @@ class Tokenizer:
      vocab = vocab + [v+'</w>' for v in vocab]
      for merge in merges:
        vocab.append(''.join(merge))
-      vocab.extend(['<|startoftext|>', '<|endoftext|>'])
+      if self.version == "sd_mlperf_v5_0":
+        import regex
+        vocab.extend(['<start_of_text>', '<end_of_text>'])
+        self.cache = {'<start_of_text>': '<start_of_text>', '<end_of_text>': '<end_of_text>'}
+        self.pat = regex.compile(r"""<start_of_text>|<end_of_text>|'s|'t|'re|'ve|'m|'ll|'d|[\p{L}]+|[\p{N}]|[^\s\p{L}\p{N}]+""", regex.IGNORECASE)
+      else:
+        vocab.extend(['<|startoftext|>', '<|endoftext|>'])
+        self.cache = {'<|startoftext|>': '<|startoftext|>', '<|endoftext|>': '<|endoftext|>'}
+        self.pat = re.compile(r"""<\|startoftext\|>|<\|endoftext\|>|'s|'t|'re|'ve|'m|'ll|'d|[^\s]+""", re.IGNORECASE)
      self.encoder = dict(zip(vocab, range(len(vocab))))
      self.bpe_ranks = dict(zip(merges, range(len(merges))))
-      self.cache = {'<|startoftext|>': '<|startoftext|>', '<|endoftext|>': '<|endoftext|>'}
-      self.pat = re.compile(r"""<\|startoftext\|>|<\|endoftext\|>|'s|'t|'re|'ve|'m|'ll|'d|[^\s]+""", re.IGNORECASE)

    def bpe(self, token):
      if token in self.cache:
@@ -110,8 +119,17 @@ class Tokenizer:

    def encode(self, text:str, pad_with_zeros:bool=False) -> List[int]:
      bpe_tokens: List[int] = []
-      text = Tokenizer.whitespace_clean(text.strip()).lower()
-      for token in re.findall(self.pat, text):
+      if self.version == "sd_mlperf_v5_0":
+        import regex, ftfy, html
+        text = ftfy.fix_text(text)
+        text = html.unescape(html.unescape(text)).strip()
+        text = Tokenizer.whitespace_clean(text).lower()
+        re_module = regex
+      else:
+        text = Tokenizer.whitespace_clean(text.strip()).lower()
+        re_module = re
+
+      for token in re_module.findall(self.pat, text):
        token = ''.join(self.byte_encoder[b] for b in token.encode('utf-8'))
        bpe_tokens.extend(self.encoder[bpe_token] for bpe_token in self.bpe(token).split(' '))
      # Truncation, keeping two slots for start and end tokens.
@@ -252,10 +270,8 @@ class Open:
      q,k,v = [y.reshape(T, B*self.n_heads, self.d_head).transpose(0, 1).reshape(B, self.n_heads, T, self.d_head) for y in proj.chunk(3)]

      attn_output = Tensor.scaled_dot_product_attention(q, k, v, attn_mask=attn_mask)
-      attn_output = attn_output.permute(2, 0, 1, 3).reshape(T*B, C)
-
+      attn_output = attn_output.permute(2, 0, 1, 3).reshape(T, B, C)
      attn_output = self.out_proj(attn_output)
-      attn_output = attn_output.reshape(T, B, C)

      return attn_output

@@ -263,9 +279,10 @@ class Open:
    def __init__(self, dims, hidden_dims):
      self.c_fc   = Linear(dims, hidden_dims)
      self.c_proj = Linear(hidden_dims, dims)
+      self.gelu = gelu

    def __call__(self, x:Tensor) -> Tensor:
-      return x.sequential([self.c_fc, Tensor.gelu, self.c_proj])
+      return x.sequential([self.c_fc, self.gelu, self.c_proj])

  # https://github.com/mlfoundations/open_clip/blob/58e4e39aaabc6040839b0d2a7e8bf20979e4558a/src/open_clip/transformer.py#L210
  class ResidualAttentionBlock:
@@ -350,15 +367,15 @@ class Open:
 # https://github.com/Stability-AI/generative-models/blob/fbdc58cab9f4ee2be7a5e1f2e2787ecd9311942f/sgm/modules/encoders/modules.py#L396
 # https://github.com/Stability-AI/generative-models/blob/fbdc58cab9f4ee2be7a5e1f2e2787ecd9311942f/sgm/modules/encoders/modules.py#L498
 class FrozenOpenClipEmbedder(Embedder):
-  def __init__(self, dims:int, n_heads:int, layers:int, return_pooled:bool, ln_penultimate:bool=False):
-    self.tokenizer = Tokenizer.ClipTokenizer()
+  def __init__(self, dims:int, n_heads:int, layers:int, return_pooled:bool, ln_penultimate:bool=False, clip_tokenizer_version=None):
+    self.tokenizer = Tokenizer.ClipTokenizer(version=clip_tokenizer_version)
    self.model = Open.ClipTextTransformer(dims, n_heads, layers)
    self.return_pooled = return_pooled
    self.input_key = "txt"
    self.ln_penultimate = ln_penultimate

  def tokenize(self, text:str, device:Optional[str]=None) -> Tensor:
-    return Tensor(self.tokenizer.encode(text, pad_with_zeros=True), dtype=dtypes.int64, device=device).reshape(1,-1)
+    return Tensor(self.tokenizer.encode(text, pad_with_zeros=True), dtype=dtypes.int32, device=device).reshape(1,-1)

  def text_transformer_forward(self, x:Tensor, attn_mask:Optional[Tensor]=None):
    for r in self.model.transformer.resblocks:
@@ -449,7 +466,7 @@ class OpenClipEncoder:
    x = x + self.positional_embedding
    x = self.transformer(x, attn_mask=self.attn_mask)
    x = self.ln_final(x)
-    x = x[:, tokens.argmax(axis=-1)]
+    x = x[Tensor.arange(x.shape[0], device=x.device), tokens.argmax(axis=-1)]
    x = x @ self.text_projection
    return x

@@ -270,8 +270,10 @@ class FidInceptionV3:
    self.Mixed_7b = inception.Mixed_7b
    self.Mixed_7c = inception.Mixed_7c

-  def load_from_pretrained(self):
-    state_dict = torch_load(str(fetch("https://github.com/mseitzer/pytorch-fid/releases/download/fid_weights/pt_inception-2015-12-05-6726825d.pth", "pt_inception-2015-12-05-6726825d.pth")))
+  def load_from_pretrained(self, path=None):
+    if path is None:
+      path = fetch("https://github.com/mseitzer/pytorch-fid/releases/download/fid_weights/pt_inception-2015-12-05-6726825d.pth", "pt_inception-2015-12-05-6726825d.pth")
+    state_dict = torch_load(str(path))
    for k,v in state_dict.items():
      if k.endswith(".num_batches_tracked"):
        state_dict[k] = v.reshape(1)
@@ -249,8 +249,5 @@ def convert_from_gguf(weights:dict[str, Tensor], n_layers:int):
  return sd

 def fix_bf16(weights:dict[Any, Tensor]):
-  if getenv("SUPPORT_BF16", 1):
-    # TODO: without casting to float16, 70B llama OOM on tinybox.
-    return {k:v.cast(dtypes.float32).cast(dtypes.float16) if v.dtype == dtypes.bfloat16 else v for k,v in weights.items()}
-  # TODO: check if device supports bf16
-  return {k:v.llvm_bf16_cast(dtypes.half).to(v.device) if v.dtype == dtypes.bfloat16 else v for k,v in weights.items()}
+  # TODO: without casting to float16, 70B llama OOM on tinybox.
+  return {k:v.cast(dtypes.float32).cast(dtypes.float16) if v.dtype == dtypes.bfloat16 else v for k,v in weights.items()}
@@ -1,21 +1,24 @@
-from tinygrad import Tensor, dtypes
-from tinygrad.nn import Linear, Conv2d, GroupNorm, LayerNorm
+from tinygrad import Tensor, dtypes, nn
 from tinygrad.device import is_dtype_supported
-from typing import Optional, Union, List, Any, Tuple
+from typing import Optional, Union, List, Any, Tuple, Callable
 import math

+# allow for monkeypatching
+Linear, Conv2d, GroupNorm, LayerNorm = nn.Linear, nn.Conv2d, nn.GroupNorm, nn.LayerNorm
+attention, gelu, mixed_precision_dtype = Tensor.scaled_dot_product_attention, Tensor.gelu, dtypes.float16
+
 # https://github.com/Stability-AI/generative-models/blob/fbdc58cab9f4ee2be7a5e1f2e2787ecd9311942f/sgm/modules/diffusionmodules/util.py#L207
 def timestep_embedding(timesteps:Tensor, dim:int, max_period=10000):
  half = dim // 2
  freqs = (-math.log(max_period) * Tensor.arange(half, device=timesteps.device) / half).exp()
  args = timesteps.unsqueeze(1) * freqs.unsqueeze(0)
  out = Tensor.cat(args.cos(), args.sin(), dim=-1)
-  return out.cast(dtypes.float16) if is_dtype_supported(dtypes.float16) else out
+  return out.cast(mixed_precision_dtype) if is_dtype_supported(mixed_precision_dtype) else out

 class ResBlock:
-  def __init__(self, channels:int, emb_channels:int, out_channels:int):
+  def __init__(self, channels:int, emb_channels:int, out_channels:int, num_groups:int=32):
    self.in_layers = [
-      GroupNorm(32, channels),
+      GroupNorm(num_groups, channels),
      Tensor.silu,
      Conv2d(channels, out_channels, 3, padding=1),
    ]
@@ -24,7 +27,7 @@ class ResBlock:
      Linear(emb_channels, out_channels),
    ]
    self.out_layers = [
-      GroupNorm(32, out_channels),
+      GroupNorm(num_groups, out_channels),
      Tensor.silu,
      lambda x: x,  # needed for weights loading code to work
      Conv2d(out_channels, out_channels, 3, padding=1),
@@ -45,35 +48,37 @@ class CrossAttention:
    self.to_v = Linear(ctx_dim,   n_heads*d_head, bias=False)
    self.num_heads = n_heads
    self.head_size = d_head
+    self.attn = attention
    self.to_out = [Linear(n_heads*d_head, query_dim)]

  def __call__(self, x:Tensor, ctx:Optional[Tensor]=None) -> Tensor:
    ctx = x if ctx is None else ctx
    q,k,v = self.to_q(x), self.to_k(ctx), self.to_v(ctx)
    q,k,v = [y.reshape(x.shape[0], -1, self.num_heads, self.head_size).transpose(1,2) for y in (q,k,v)]
-    attention = Tensor.scaled_dot_product_attention(q, k, v).transpose(1,2)
+    attention = self.attn(q, k, v).transpose(1,2)
    h_ = attention.reshape(x.shape[0], -1, self.num_heads * self.head_size)
    return h_.sequential(self.to_out)

 class GEGLU:
  def __init__(self, dim_in:int, dim_out:int):
    self.proj = Linear(dim_in, dim_out * 2)
+    self.gelu = gelu
    self.dim_out = dim_out

  def __call__(self, x:Tensor) -> Tensor:
    x, gate = self.proj(x).chunk(2, dim=-1)
-    return x * gate.gelu()
+    return x * self.gelu(gate)

 class FeedForward:
  def __init__(self, dim:int, mult:int=4):
-    self.net = [
+    self.net: tuple[GEGLU, Callable, nn.Linear] = (
      GEGLU(dim, dim*mult),
      lambda x: x,  # needed for weights loading code to work
      Linear(dim*mult, dim)
-    ]
+    )

  def __call__(self, x:Tensor) -> Tensor:
-    return x.sequential(self.net)
+    return x.sequential(list(self.net))

 class BasicTransformerBlock:
  def __init__(self, dim:int, ctx_dim:int, n_heads:int, d_head:int):
@@ -92,12 +97,13 @@ class BasicTransformerBlock:

 # https://github.com/Stability-AI/generative-models/blob/fbdc58cab9f4ee2be7a5e1f2e2787ecd9311942f/sgm/modules/attention.py#L619
 class SpatialTransformer:
-  def __init__(self, channels:int, n_heads:int, d_head:int, ctx_dim:Union[int,List[int]], use_linear:bool, depth:int=1):
+  def __init__(self, channels:int, n_heads:int, d_head:int, ctx_dim:Union[int,List[int]], use_linear:bool, depth:int=1,
+               norm_eps:float=1e-5):
    if isinstance(ctx_dim, int):
      ctx_dim = [ctx_dim]*depth
    else:
      assert isinstance(ctx_dim, list) and depth == len(ctx_dim)
-    self.norm = GroupNorm(32, channels)
+    self.norm = GroupNorm(32, channels, eps=norm_eps)
    assert channels == n_heads * d_head
    self.proj_in  = Linear(channels, channels) if use_linear else Conv2d(channels, channels, 1)
    self.transformer_blocks = [BasicTransformerBlock(channels, ctx_dim[d], n_heads, d_head) for d in range(depth)]
@@ -134,7 +140,9 @@ class Upsample:

 # https://github.com/Stability-AI/generative-models/blob/fbdc58cab9f4ee2be7a5e1f2e2787ecd9311942f/sgm/modules/diffusionmodules/openaimodel.py#L472
 class UNetModel:
-  def __init__(self, adm_in_ch:Optional[int], in_ch:int, out_ch:int, model_ch:int, attention_resolutions:List[int], num_res_blocks:int, channel_mult:List[int], transformer_depth:List[int], ctx_dim:Union[int,List[int]], use_linear:bool=False, d_head:Optional[int]=None, n_heads:Optional[int]=None):
+  def __init__(self, adm_in_ch:Optional[int], in_ch:int, out_ch:int, model_ch:int, attention_resolutions:List[int], num_res_blocks:int,
+               channel_mult:List[int], transformer_depth:List[int], ctx_dim:Union[int,List[int]], use_linear:bool=False, d_head:Optional[int]=None,
+               n_heads:Optional[int]=None, num_groups:int=32, st_norm_eps:float=1e-5):
    self.model_ch = model_ch
    self.num_res_blocks = [num_res_blocks] * len(channel_mult)

@@ -174,12 +182,12 @@ class UNetModel:
    for idx, mult in enumerate(channel_mult):
      for _ in range(self.num_res_blocks[idx]):
        layers: List[Any] = [
-          ResBlock(ch, time_embed_dim, model_ch*mult),
+          ResBlock(ch, time_embed_dim, model_ch*mult, num_groups),
        ]
        ch = mult * model_ch
        if ds in attention_resolutions:
          d_head, n_heads = get_d_and_n_heads(ch)
-          layers.append(SpatialTransformer(ch, n_heads, d_head, ctx_dim, use_linear, depth=transformer_depth[idx]))
+          layers.append(SpatialTransformer(ch, n_heads, d_head, ctx_dim, use_linear, depth=transformer_depth[idx], norm_eps=st_norm_eps))

        self.input_blocks.append(layers)
        input_block_channels.append(ch)
@@ -193,9 +201,9 @@ class UNetModel:

    d_head, n_heads = get_d_and_n_heads(ch)
    self.middle_block: List = [
-      ResBlock(ch, time_embed_dim, ch),
-      SpatialTransformer(ch, n_heads, d_head, ctx_dim, use_linear, depth=transformer_depth[-1]),
-      ResBlock(ch, time_embed_dim, ch),
+      ResBlock(ch, time_embed_dim, ch, num_groups),
+      SpatialTransformer(ch, n_heads, d_head, ctx_dim, use_linear, depth=transformer_depth[-1], norm_eps=st_norm_eps),
+      ResBlock(ch, time_embed_dim, ch, num_groups),
    ]

    self.output_blocks = []
@@ -203,13 +211,13 @@ class UNetModel:
      for i in range(self.num_res_blocks[idx] + 1):
        ich = input_block_channels.pop()
        layers = [
-          ResBlock(ch + ich, time_embed_dim, model_ch*mult),
+          ResBlock(ch + ich, time_embed_dim, model_ch*mult, num_groups),
        ]
        ch = model_ch * mult

        if ds in attention_resolutions:
          d_head, n_heads = get_d_and_n_heads(ch)
-          layers.append(SpatialTransformer(ch, n_heads, d_head, ctx_dim, use_linear, depth=transformer_depth[idx]))
+          layers.append(SpatialTransformer(ch, n_heads, d_head, ctx_dim, use_linear, depth=transformer_depth[idx], norm_eps=st_norm_eps))

        if idx > 0 and i == self.num_res_blocks[idx]:
          layers.append(Upsample(ch))
@@ -217,7 +225,7 @@ class UNetModel:
        self.output_blocks.append(layers)

    self.out = [
-      GroupNorm(32, ch),
+      GroupNorm(num_groups, ch),
      Tensor.silu,
      Conv2d(model_ch, out_ch, 3, padding=1),
    ]
@@ -230,10 +238,10 @@ class UNetModel:
      assert y.shape[0] == x.shape[0]
      emb = emb + y.sequential(self.label_emb[0])

-    if is_dtype_supported(dtypes.float16):
-      emb = emb.cast(dtypes.float16)
-      ctx = ctx.cast(dtypes.float16)
-      x   = x  .cast(dtypes.float16)
+    if is_dtype_supported(mixed_precision_dtype):
+      emb = emb.cast(mixed_precision_dtype)
+      ctx = ctx.cast(mixed_precision_dtype)
+      x   = x  .cast(mixed_precision_dtype)

    def run(x:Tensor, bb) -> Tensor:
      if isinstance(bb, ResBlock): x = bb(x, emb)
@@ -272,4 +272,4 @@ def compare_launch_state(states, good_states):

  return True, "PASS"

-# IOCTL=1 PTX=1 CUDA=1 python3 test/test_ops.py TestOps.test_tiny_add
+# IOCTL=1 CUDA=1 CUDA_PTX=1 python3 test/test_ops.py TestOps.test_tiny_add
@@ -1,7 +1,6 @@
 from tinygrad import Tensor
 from tinygrad.tensor import _to_np_dtype
-from tinygrad.frontend.onnx import OnnxRunner
-from extra.onnx import OnnxValue
+from tinygrad.nn.onnx import OnnxRunner, OnnxValue
 import numpy as np
 import onnxruntime as ort

@@ -7,7 +7,7 @@ rm $LOGOPS
 test/external/process_replay/reset.py

 CI=1 python3 -m pytest -n=auto test/test_ops.py test/test_nn.py test/test_winograd.py test/models/test_real_world.py --durations=20
-GPU=1 python3 -m pytest test/test_tiny.py
+CL=1 python3 -m pytest test/test_tiny.py

 # extract, sort and uniq
 extra/optimization/extract_dataset.py
@@ -1,6 +1,6 @@
 # stuff needed to unpack a kernel
 from tinygrad import Variable
-from tinygrad.codegen.opt.kernel import Opt, OptOps
+from tinygrad.codegen.opt import Opt, OptOps
 from tinygrad.uop.ops import UOp, Ops, KernelInfo
 from tinygrad.dtype import dtypes, PtrDType
 from tinygrad.shape.shapetracker import ShapeTracker
@@ -115,7 +115,7 @@ def time_linearizer(lin:Kernel, rawbufs:list[Buffer], allow_test_size=True, max_
  assert dev.compiler is not None

  rawbufs = _ensure_buffer_alloc(rawbufs)
-  var_vals: dict[Variable, int] = {k:int(k.vmax+k.vmin)//2 for k in lin.ast.variables()}
+  var_vals: dict[str, int] = {k.expr:int(k.vmax+k.vmin)//2 for k in lin.ast.variables()}
  p = get_program(lin.get_optimized_ast(), lin.opts)
  tms = _time_program(p, dev.compiler.compile(p.src), var_vals, rawbufs,
                      max_global_size=max_global_size if allow_test_size else None, clear_l2=clear_l2, cnt=cnt, name=to_function_name(lin.name))
@@ -16,9 +16,9 @@ class TestBeamSearch(unittest.TestCase):
    BEAM.value = self.old_beam

  def test_variable_ast_beam(self):
-    with Context(IGNORE_OOB=1):
-      a = rand(3, 3).reshape((Variable("a", 1, 10).bind(3), 3))
-      a = (a+1).realize()
+    vi = Variable("a", 1, 10).bind(3)
+    a = rand(10, 3)[:vi]
+    a = (a+1).realize()

  def test_big_prime_number(self):
    a = rand(367, 367)
@@ -42,18 +42,16 @@ class TestBeamSearch(unittest.TestCase):

  def test_variable_big_prime_number(self):
    v = Variable("v", 1, 400).bind(367)
-    a = rand(367, 367)
-    b = rand(367, 367)
-    with Context(IGNORE_OOB=1):
-      c = (a.reshape(367, v) @ b.reshape(v, 367)).realize()
-      np.testing.assert_allclose(c.numpy(), a.numpy() @ b.numpy(), atol=1e-4, rtol=1e-4)
+    a = rand(367, 400)
+    b = rand(400, 367)
+    c = (a[:, :v] @ b[:v, :]).realize()
+    np.testing.assert_allclose(c.numpy(), a[:, :367].numpy() @ b[:367, :].numpy(), atol=1e-4, rtol=1e-4)

  def test_variable_shrink_prime_number(self):
    v = Variable("v", 1, 400).bind(367)
    a = rand(400, 367)
-    with Context(IGNORE_OOB=1):
-      b = (a.shrink(((0,v), None))+1).reshape(367,367).realize()
-      np.testing.assert_allclose(b.numpy(), a.numpy()[:367]+1, atol=1e-4, rtol=1e-4)
+    b = (a.shrink(((0,v), None))+1)[:367,:367].realize()
+    np.testing.assert_allclose(b.numpy(), a.numpy()[:367]+1, atol=1e-4, rtol=1e-4)

  def test_no_mutate_rawbuffers(self):
    a = rand(3, 3).realize()
@@ -1,6 +1,6 @@
 import ctypes, array
 from hexdump import hexdump
-from tinygrad.runtime.ops_gpu import GPUDevice
+from tinygrad.runtime.ops_cl import CLDevice
 from tinygrad.helpers import getenv, to_mv, mv_address
 from tinygrad.dtype import dtypes
 from tinygrad import Tensor, TinyJit
@@ -8,7 +8,7 @@ from tinygrad.runtime.autogen import opencl as cl
 if getenv("IOCTL"): import extra.qcom_gpu_driver.opencl_ioctl  # noqa: F401  # pylint: disable=unused-import

 # create raw opencl buffer.
-gdev = GPUDevice()
+gdev = CLDevice()
 cl_buf = cl.clCreateBuffer(gdev.context, cl.CL_MEM_READ_WRITE, 0x100, None, status := ctypes.c_int32())
 assert status.value == 0

@@ -673,6 +673,7 @@ impl<'a> Thread<'a> {
                                        39 => f32::log2(s0),
                                        42 => 1.0 / s0,
                                        43 => 1.0 / s0,
+                                        46 => 1.0 / f32::sqrt(s0),
                                        51 => f32::sqrt(s0),
                                        _ => todo_instr!(instruction)?,
                                    }
@@ -929,7 +930,7 @@ impl<'a> Thread<'a> {

            let op = ((instr >> 16) & 0x3ff) as u32;
            match op {
-                764 | 765 | 288 | 289 | 290 | 766 | 768 | 769 => {
+                764 | 765 | 288 | 289 | 290 | 766 | 767 | 768 | 769 => {
                    let vdst = (instr & 0xff) as usize;
                    let sdst = ((instr >> 8) & 0x7f) as usize;
                    let f = |i: u32| -> usize { ((instr >> i) & 0x1ff) as usize };
@@ -943,6 +944,16 @@ impl<'a> Thread<'a> {
                    assert_eq!(clmp, 0);

                    let vcc = match op {
+                        767 => {
+                            let (s0, s1, s2): (u32, u32, u64) = (self.val(s0), self.val(s1), self.val(s2));
+                            let (mul_result, overflow_mul) = (s0 as i64).overflowing_mul(s1 as i64);
+                            let (ret, overflow_add) = mul_result.overflowing_add(s2 as i64);
+                            let overflowed = overflow_mul || overflow_add;
+                            if self.exec.read() {
+                                self.vec_reg.write64(vdst, ret as u64);
+                            }
+                            overflowed
+                        },
                        766 => {
                            let (s0, s1, s2): (u32, u32, u64) = (self.val(s0), self.val(s1), self.val(s2));
                            let (mul_result, overflow_mul) = (s0 as u64).overflowing_mul(s1 as u64);
@@ -1246,7 +1257,7 @@ impl<'a> Thread<'a> {
                            }

                            let ret = match op {
-                                257 | 259 | 299 | 260 | 261 | 264 | 272 | 392 | 426 | 531 | 537 | 540 | 551 | 567 | 796 => {
+                                257 | 259 | 299 | 260 | 261 | 264 | 272 | 392 | 426 | 430 | 531 | 537 | 540 | 551 | 567 | 796 => {
                                    let s0 = f32::from_bits(s0).negate(0, neg).absolute(0, abs);
                                    let s1 = f32::from_bits(s1).negate(1, neg).absolute(1, abs);
                                    let s2 = f32::from_bits(s2).negate(2, neg).absolute(2, abs);
@@ -1258,6 +1269,7 @@ impl<'a> Thread<'a> {
                                        272 => f32::max(s0, s1),
                                        299 => f32::mul_add(s0, s1, f32::from_bits(self.vec_reg[vdst])),
                                        426 => s0.recip(),
+                                        430 => 1.0 / f32::sqrt(s0),
                                        531 => f32::mul_add(s0, s1, s2),
                                        537 => f32::min(f32::min(s0, s1), s2),
                                        540 => f32::max(f32::max(s0, s1), s2),
@@ -2625,6 +2637,14 @@ mod test_vop1 {
        assert_eq!(thread.vec_reg[3], 1071644672);
    }

+    #[test]
+    fn test_v_rsq_f32() {
+        let mut thread = _helper_test_thread();
+        thread.vec_reg[0] = f32::to_bits(4.0);
+        r(&vec![0x7E005D00, END_PRG], &mut thread);
+        assert_eq!(f32::from_bits(thread.vec_reg[0]), 0.5);
+    }
+
    #[test]
    fn test_v_frexp_exp_i32_f64() {
        [(3573412790272.0, 42), (69.0, 7), (2.0, 2), (f64::NEG_INFINITY, 0)]
@@ -58,7 +58,7 @@ if __name__ == "__main__":
            GlobalCounters.kernel_count -= 1

        if not getenv("NOOPT"): k.apply_opts(hand_coded_optimizations(k))
-        p2 = get_program(k.get_optimized_ast(), k.opts)
+        p2 = get_program(k.ast, k.opts, k.applied_opts)
        new_ei = replace(ei, prg=CompiledRunner(p2))
        new_ei.run()
        new_jit.append(new_ei)
@@ -4,7 +4,7 @@

 Only supported on 7900XTX, requires either AM (`rmmod amdgpu`) or disabling power gating on AMD (`ppfeaturemask=0xffff3fff`, don't forget to rebuild initramfs)

-SQTT is implemented on top of normal tinygrad PROFILE=1, `PROFILE=1 SQTT=1` to get profile pickle with sqtt data embedded in it.
+SQTT is implemented on top of normal tinygrad profiling, `VIZ=1 SQTT=1` to get profile pickle with sqtt data embedded in it.

 `SQTT_BUFFER_SIZE=X` to change size of SQTT buffer (per shader engine, 6 SEs on 7900xtx) in megabytes, default 256.

@@ -0,0 +1,20 @@
+from extra.optimization.helpers import load_worlds, ast_str_to_ast
+from tinygrad.helpers import tqdm
+from tinygrad.uop.ops import pyrender, UOp, Ops
+from tinygrad import dtypes
+from tinygrad.shape.shapetracker import ShapeTracker, View
+inf, nan = float('inf'), float('nan')
+
+if __name__ == "__main__":
+  ast_strs = load_worlds()
+  for i, ast_str in enumerate(tqdm(ast_strs)):
+    good_ast = ast_str_to_ast(ast_str)
+    code = '\n'.join(pyrender(good_ast))
+    print("\n***************\n\n"+code)
+    exec(code)
+    if str(good_ast) != str(ast):
+      print(code)
+      print("MISMATCH")
+      print(good_ast)
+      print(ast)
+      break
@@ -4,13 +4,13 @@ import struct
 import json
 import traceback
 import numpy as np
-from tinygrad.runtime.ops_gpu import CLProgram, compile_gpu
+from tinygrad.runtime.ops_cl import CLProgram, compile_gpu
 from tinygrad.device import Device
 from tinygrad.helpers import DEBUG, getenv
 from collections import defaultdict
 import pyopencl as cl
-from tinygrad.runtime.ops_gpu import OSX_TIMING_RATIO
-CL = Device["GPU"]
+from tinygrad.runtime.ops_cl import OSX_TIMING_RATIO
+CL = Device["CL"]

 DEBUGCL = getenv("DEBUGCL", 0)
 FLOAT16 = getenv("FLOAT16", 0)
@@ -110,7 +110,7 @@ class Thneed:
    prgs = {}
    for o in jdat['binaries']:
      nptr = ptr + o['length']
-      prgs[o['name']] = CLProgram(Device["GPU"], o['name'], weights[ptr:nptr])
+      prgs[o['name']] = CLProgram(Device["CL"], o['name'], weights[ptr:nptr])
      ptr = nptr

    # populate the cl_cache
@@ -267,7 +267,7 @@ class Thneed:
    for prg, args in self.cl_cache:
      events.append(prg.clprg(CL.queue, *args))
    mt = time.monotonic()
-    Device["GPU"].synchronize()
+    Device["CL"].synchronize()
    et = time.monotonic() - st
    print(f"submit in {(mt-st)*1000.0:.2f} ms, total runtime is {et*1000.0:.2f} ms")

@@ -0,0 +1,74 @@
+# include directory copied from https://github.com/HazyResearch/ThunderMittens
+# https://hazyresearch.stanford.edu/blog/2024-11-28-tk-mlx
+
+gemm = """
+#include <metal_stdlib>
+#include "include/tk.metal"
+using namespace mittens;
+
+#define GEMM_PARAMS_DEF(T) \
+    device T* D [[buffer(0)]], \
+    device T* A [[buffer(1)]], \
+    device T* B [[buffer(2)]], \
+    const constant int &N [[buffer(3)]], \
+    const constant int &K [[buffer(4)]], \
+    const constant int &M [[buffer(5)]], \
+    uint3 tg_id [[threadgroup_position_in_grid]], \
+    uint simd_lane_id [[thread_index_in_simdgroup]]
+
+template<typename T, unsigned N_BLOCK, unsigned K_BLOCK, unsigned M_BLOCK>
+kernel void matmul_naive(GEMM_PARAMS_DEF(T)) {
+  using global_layout = gl<T, 1, 1, -1, -1>;
+  global_layout gl_a(A, nullptr, nullptr, N, K);
+  global_layout gl_b(B, nullptr, nullptr, K, M);
+  global_layout gl_d(D, nullptr, nullptr, N, M);
+  rt<T,     N_BLOCK * TILE_DIM, K_BLOCK * TILE_DIM> a_reg;
+  rt<T,     K_BLOCK * TILE_DIM, M_BLOCK * TILE_DIM> b_reg;
+  rt<float, N_BLOCK * TILE_DIM, M_BLOCK * TILE_DIM> d_reg;
+  zero(d_reg);
+  #pragma clang loop unroll(full)
+  for (int k = 0; k < K / (K_BLOCK * TILE_DIM); k++) {
+    load(a_reg, gl_a, {0, 0, (int)tg_id.y, k}, simd_lane_id);
+    load(b_reg, gl_b, {0, 0, k, (int)tg_id.x}, simd_lane_id);
+    mma_AB(d_reg, a_reg, b_reg, d_reg);
+  }
+  store(gl_d, d_reg, {0, 0, (int)tg_id.y, (int)tg_id.x}, simd_lane_id);
+}
+
+#define instantiate_matmul_custom(type_name, T) \
+   template [[host_name("matmul_custom_" #type_name)]] [[kernel]] \
+   void matmul_naive<T, 4, 2, 4>(GEMM_PARAMS_DEF(T)); \
+
+instantiate_matmul_custom(float32, float);
+"""
+
+from tinygrad import Device, Tensor, Context
+
+if __name__ == "__main__":
+  device = Device["METAL"]
+  lib = device.compiler.compile(gemm)
+  prg = device.runtime("matmul_custom_float32", lib)
+
+  N = 4096
+  a = Tensor.randn(N, N)
+  b = Tensor.randn(N, N)
+  c = Tensor.empty(N, N)
+  Tensor.realize(a, b, c)
+
+  TILE_DIM = 8
+  N_BLOCK = 4
+  M_BLOCK = 4
+
+  gsz = (N // (M_BLOCK * TILE_DIM), N // (N_BLOCK * TILE_DIM), 1)
+  for _ in range(5):
+    et = prg(c.uop.buffer.ensure_allocated()._buf, a.uop.buffer._buf, b.uop.buffer._buf,
+            global_size=gsz, local_size=(32,1,1), vals=(N, N, N), wait=True)
+    print(f"{N*N*N*2/(et*1e9):2f} GFLOPS")
+
+  for _ in range(5):
+    with Context(DEBUG=2):
+      ref = (a@b).realize()
+
+  print((ref-c).mean().item())
+
+
@@ -0,0 +1,392 @@
+/**
+ * @file
+ * @brief Basic operations on generic types.
+ */
+#pragma once
+#include "base_types.metal"
+#include <metal_math>
+
+namespace mittens {
+/**
+ * @namespace base_ops
+ *
+ * @brief A namespace for operations on basic data types.
+ */
+namespace base_ops {
+#define TEMPLATE_OPS_SINGLE(func_contents) \
+    template<typename T> static METAL_FUNC T op(device const T &x)      { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &x) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread const T &x)      { func_contents }
+
+#define TEMPLATE_OPS_OVERRIDE_SINGLE(T, op_name, func_contents) \
+    template<> METAL_FUNC T op_name::op<T>(device const T &x)      { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &x) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread const T &x)      { func_contents }
+
+#define TEMPLATE_OPS_DOUBLE(func_contents) \
+    template<typename T> static METAL_FUNC T op(device      const T &a, device      const T &b) { func_contents } \
+    template<typename T> static METAL_FUNC T op(device      const T &a, threadgroup const T &b) { func_contents } \
+    template<typename T> static METAL_FUNC T op(device      const T &a, thread      const T &b) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, device      const T &b) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, threadgroup const T &b) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, thread      const T &b) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread      const T &a, device      const T &b) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread      const T &a, threadgroup const T &b) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread      const T &a, thread      const T &b) { func_contents }
+
+#define TEMPLATE_OPS_OVERRIDE_DOUBLE(T, op_name, func_contents) \
+    template<> METAL_FUNC T op_name::op<T>(device      const T &a, device      const T &b) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(device      const T &a, threadgroup const T &b) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(device      const T &a, thread      const T &b) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, device      const T &b) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, threadgroup const T &b) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, thread      const T &b) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, device      const T &b) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, threadgroup const T &b) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, thread      const T &b) { func_contents }
+    
+#define TEMPLATE_OPS_TRIPLE(func_contents) \
+    template<typename T> static METAL_FUNC T op(device      const T &a, device      const T &b, device      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(device      const T &a, device      const T &b, threadgroup const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(device 	    const T &a, device      const T &b, thread      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(device      const T &a, threadgroup const T &b, device      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(device 	    const T &a, threadgroup const T &b, threadgroup const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(device      const T &a, threadgroup const T &b, thread      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(device      const T &a, thread      const T &b, device      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(device      const T &a, thread      const T &b, threadgroup const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(device      const T &a, thread      const T &b, thread      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, device      const T &b, device      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, device      const T &b, threadgroup const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, device      const T &b, thread      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, threadgroup const T &b, device      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, threadgroup const T &b, threadgroup const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, threadgroup const T &b, thread      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, thread      const T &b, device      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, thread      const T &b, threadgroup const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &a, thread      const T &b, thread      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread      const T &a, device      const T &b, device 	    const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread 	    const T &a, device      const T &b, threadgroup const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread      const T &a, device      const T &b, thread      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread      const T &a, threadgroup const T &b, device      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread      const T &a, threadgroup const T &b, threadgroup const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread      const T &a, threadgroup const T &b, thread      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread      const T &a, thread      const T &b, device      const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread      const T &a, thread      const T &b, threadgroup const T &c) { func_contents } \
+    template<typename T> static METAL_FUNC T op(thread      const T &a, thread      const T &b, thread      const T &c) { func_contents }
+
+#define TEMPLATE_OPS_OVERRIDE_TRIPLE(T, op_name, func_contents) \
+    template<> METAL_FUNC T op_name::op<T>(device      const T &a, device      const T &b, device      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(device      const T &a, device      const T &b, threadgroup const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(device      const T &a, device      const T &b, thread      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(device      const T &a, threadgroup const T &b, device      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(device 	   const T &a, threadgroup const T &b, threadgroup const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(device      const T &a, threadgroup const T &b, thread      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(device      const T &a, thread      const T &b, device      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(device      const T &a, thread      const T &b, threadgroup const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(device      const T &a, thread      const T &b, thread      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, device      const T &b, device      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, device      const T &b, threadgroup const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, device      const T &b, thread      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, threadgroup const T &b, device      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, threadgroup const T &b, threadgroup const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, threadgroup const T &b, thread      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, thread      const T &b, device      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, thread      const T &b, threadgroup const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(threadgroup const T &a, thread      const T &b, thread      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, device      const T &b, device      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, device      const T &b, threadgroup const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, device      const T &b, thread      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, threadgroup const T &b, device      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, threadgroup const T &b, threadgroup const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, threadgroup const T &b, thread      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, thread      const T &b, device      const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, thread      const T &b, threadgroup const T &c) { func_contents } \
+    template<> METAL_FUNC T op_name::op<T>(thread      const T &a, thread      const T &b, thread      const T &c) { func_contents }
+
+
+
+/* ----------  CONST OPS  ---------- */
+
+/**
+ * @brief Represents the zero constant operation.
+ *
+ * This operation returns the zero value of the specified type.
+ *
+ * @tparam T The data type for which to return the zero value.
+ * @return The zero value of type T.
+ */
+struct zero {
+    template<typename T, typename... args> static METAL_FUNC constexpr T op(args... _) { return base_types::constants<T>::zero();      }
+};
+/**
+ * @brief Represents the one constant operation.
+ *
+ * This operation returns the one value of the specified type.
+ *
+ * @tparam T The data type for which to return the one value.
+ * @return The one value of type T.
+ */
+struct one {
+    template<typename T, typename... args> static METAL_FUNC constexpr T op(args... _) { return base_types::constants<T>::one();       }
+};
+    
+/**
+ * @brief Represents the positive infinity constant operation.
+ *
+ * This operation returns the positive infinity value of the specified type.
+ *
+ * @tparam T The data type for which to return the positive infinity value.
+ * @return The positive infinity value of type T.
+ */
+struct pos_infty {
+    template<typename T, typename... args> static METAL_FUNC constexpr T op(args... _) { return base_types::constants<T>::pos_infty(); }
+};
+/**
+ * @brief Represents the negative infinity constant operation.
+ *
+ * This operation returns the negative infinity value of the specified type.
+ *
+ * @tparam T The data type for which to return the negative infinity value.
+ * @return The negative infinity value of type T.
+ */
+struct neg_infty {
+    template<typename T, typename... args> static METAL_FUNC constexpr T op(args... _) { return base_types::constants<T>::neg_infty(); }
+};
+    
+
+/* ----------  UNARY OPS  ---------- */
+/**
+ * @brief Exponential function operation.
+ *
+ * This operation calculates the exponential of the input value.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param x[in] The input value.
+ * @return The exponential of the input value.
+ */
+struct exp {
+    TEMPLATE_OPS_SINGLE(return metal::exp(x);)
+};
+
+TEMPLATE_OPS_OVERRIDE_SINGLE(bf16, exp, return bf16(metal::exp((float)x));)
+TEMPLATE_OPS_OVERRIDE_SINGLE(bf16_2, exp, return bf16_2(metal::exp(float2(x)));)
+    
+    /**
+ * @brief Exponential function operation, in base 2
+ *
+ * This operation calculates the exponential of the input value, in base 2.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param x[in] The input value.
+ * @return The exponential of the input value.
+ */
+struct exp2 {
+    template<typename T> static METAL_FUNC T op(device const T &x)      { return metal::exp2(x); } \
+    template<typename T> static METAL_FUNC T op(threadgroup const T &x) { return metal::exp2(x); } \
+    template<typename T> static METAL_FUNC T op(thread const T &x)      { return metal::exp2(x); }
+};
+    
+//template<> METAL_FUNC bf16 exp2::op<bf16>(device const bf16 &x)      { return bf16(metal::exp2(x)); } \
+//template<> METAL_FUNC bf16 exp2::op<bf16>(threadgroup const bf16 &x) { return bf16(metal::exp2(x)); } \
+//template<> METAL_FUNC bf16 exp2::op<bf16>(thread const bf16 &x)      { return bf16(metal::exp2(x)); }
+TEMPLATE_OPS_OVERRIDE_SINGLE(bf16, exp2, return bf16(metal::exp2(x));)
+TEMPLATE_OPS_OVERRIDE_SINGLE(bf16_2, exp2, return bf16_2(metal::exp2((float2)x));)
+
+/**
+ * @brief Natural log function operation.
+ *
+ * This operation calculates the natural logarithm of the input value.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param x[in] The input value.
+ * @return The natural logarithm of the input value.
+ */
+struct log {
+    TEMPLATE_OPS_SINGLE(return metal::log(x);)
+};
+TEMPLATE_OPS_OVERRIDE_SINGLE(bf16, log, return bf16(metal::log(x));)
+TEMPLATE_OPS_OVERRIDE_SINGLE(bf16_2, log, return bf16_2(metal::log((float2)x));)
+
+/**
+ * @brief Absolute value operation.
+ *
+ * This operation calculates the absolute value of the input.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param x[in] The input value.
+ * @return The absolute value of the input.
+ */
+struct abs {
+    TEMPLATE_OPS_SINGLE(return metal::abs(x);)
+};
+TEMPLATE_OPS_OVERRIDE_SINGLE(bf16  , abs, return bf16(metal::abs((float)x));)
+TEMPLATE_OPS_OVERRIDE_SINGLE(bf16_2, abs, return bf16_2(metal::abs((float2)x));)
+/**
+ * @brief Rectified Linear Unit (ReLU) operation.
+ *
+ * This operation applies the ReLU function to the input, which is the
+ * maximum of zero and the input value.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param x[in] The input value.
+ * @return The result of ReLU function applied to the input.
+ */
+struct relu {
+    TEMPLATE_OPS_SINGLE(return max(x, base_types::constants<T>::zero());)
+};
+TEMPLATE_OPS_OVERRIDE_SINGLE(bf16  , relu, return bf16(metal::max((float)x, base_types::constants<float>::zero()));)
+TEMPLATE_OPS_OVERRIDE_SINGLE(bf16_2, relu, return bf16_2(metal::max((float2)x, base_types::constants<float2>::zero()));)
+/**
+ * @brief Copy operation.
+ *
+ * This operation returns the input value unchanged.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param a[in] The input value.
+ * @return The same value as the input.
+ */
+struct copy { // for non-compile-time setters.
+    TEMPLATE_OPS_SINGLE(return x;)
+};
+
+/* ----------  BINARY OPS  ---------- */
+  
+    
+/**
+ * @brief Copy2 operation.
+ *
+ * This operation returns the second input value unchanged.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param a[in] The first input value (ignored).
+ * @param b[in] The second input value.
+ * @return The same value as the second input.
+ */
+struct copy2 { // this turns out to be a slightly hacky op that makes some code cleaner :/
+    TEMPLATE_OPS_DOUBLE(return b;)
+};
+/**
+ * @brief Sum operation.
+ *
+ * This operation calculates the sum of two input values.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param a[in] The first input value.
+ * @param b[in] The second input value.
+ * @return The sum of the input values.
+ */
+struct sum {
+    TEMPLATE_OPS_DOUBLE(return a+b;)
+};
+    
+/**
+ * @brief Subtraction operation.
+ *
+ * This operation calculates the difference between two input values.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param a[in] The first input value.
+ * @param b[in] The second input value.
+ * @return The difference between the input values.
+ */
+struct sub {
+    TEMPLATE_OPS_DOUBLE(return a-b;)
+};
+/**
+ * @brief Multiplication operation.
+ *
+ * This operation calculates the product of two input values.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param a[in] The first input value.
+ * @param b[in] The second input value.
+ * @return The product of the input values.
+ */
+struct mul {
+    TEMPLATE_OPS_DOUBLE(return a*b;)
+};
+/**
+ * @brief Division operation.
+ *
+ * This operation calculates the quotient of two input values.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param a[in] The first input value.
+ * @param b[in] The second input value.
+ * @return The quotient of the input values.
+ */
+struct div {
+    TEMPLATE_OPS_DOUBLE(return a/b;)
+};
+/**
+ * @brief Maximum operation.
+ *
+ * This operation calculates the maximum of two input values.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param a[in] The first input value.
+ * @param b[in] The second input value.
+ * @return The maximum of the input values.
+ */
+struct max {
+    TEMPLATE_OPS_DOUBLE(return metal::max(a,b);)
+};
+TEMPLATE_OPS_OVERRIDE_DOUBLE(bf16  , max, return (bf16)metal::max((float)a, (float)b);)
+TEMPLATE_OPS_OVERRIDE_DOUBLE(bf16_2, max, return (bf16_2)metal::max((float2)a, (float2)b);)
+/**
+ * @brief Minimum operation.
+ *
+ * This operation calculates the minimum of two input values.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param a[in] The first input value.
+ * @param b[in] The second input value.
+ * @return The minimum of the input values.
+ */
+struct min {
+    TEMPLATE_OPS_DOUBLE(return metal::min(a,b);)
+};
+TEMPLATE_OPS_OVERRIDE_DOUBLE(bf16  , min, return (bf16)metal::min((float)a, (float)b);)
+TEMPLATE_OPS_OVERRIDE_DOUBLE(bf16_2, min, return (bf16_2)metal::min((float2)a, (float2)b);)
+
+
+/* ----------  TERNARY OPS  ---------- */
+/**
+ * @brief Fused multiply-add operation A * B + C.
+ *
+ * This operation performs a fused multiply-add, computing (A * B) + C with only one rounding.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param a[in] The first input value.
+ * @param b[in] The second input value.
+ * @param c[in] The third input value to be added.
+ * @return The result of the fused multiply-add operation.
+ */
+struct fma_AxBtC {
+    TEMPLATE_OPS_TRIPLE(return sum::op<T>(mul::op<T>(a, b), c);)
+};
+    
+/**
+ * @brief Fused multiply-add operation A * C + B.
+ *
+ * This operation performs a fused multiply-add, computing (A * C) + B with only one rounding.
+ * This is particularly useful for attention mechanisms in neural networks.
+ *
+ * @tparam T The data type of the input and output values.
+ * @param a[in] The first input value. 
+ * @param b[in] The third input value to be added.
+ * @param c[in] The second input value.
+ * @return The result of the fused multiply-add operation.
+ */
+struct fma_AxCtB { // this is the one needed for attention
+    TEMPLATE_OPS_TRIPLE(return sum::op<T>(mul::op<T>(a, c), b);)
+};
+
+#undef TEMPLATE_OPS_SINGLE
+#undef TEMPLATE_OPS_OVERRIDE_SINGLE
+#undef TEMPLATE_OPS_DOUBLE
+#undef TEMPLATE_OPS_OVERRIDE_DOUBLE
+#undef TEMPLATE_OPS_TRIPLE
+#undef TEMPLATE_OPS_OVERRIDE_TRIPLE
+} // base_ops
+} // mittens
@@ -0,0 +1,321 @@
+
+#pragma once
+
+namespace mittens {
+    
+using bf16 = bfloat;
+using bf16_2 = bfloat2;
+using bf16_4 = bfloat4;
+//using half_2 = half2;
+     
+namespace ducks {
+namespace base_types {
+template <typename T>
+static METAL_FUNC constexpr const bool isT1() {
+    return metal::is_same<typename T::dtype, float>::value ||
+           metal::is_same<typename T::dtype, bf16 >::value ||
+           metal::is_same<typename T::dtype, half>::value;
+}
+template <typename T>
+static METAL_FUNC constexpr const bool isT2() {
+    return metal::is_same<typename T::dtype, float2>::value ||
+           metal::is_same<typename T::dtype, bf16_2>::value ||
+           metal::is_same<typename T::dtype, half2>::value;
+}
+    
+template <typename T>
+static METAL_FUNC constexpr const bool isT1Type() {
+    return metal::is_same<T, float>::value ||
+           metal::is_same<T, bf16 >::value ||
+           metal::is_same<T, half>::value;
+}
+template <typename T>
+static METAL_FUNC constexpr const bool isT2Type() {
+    return metal::is_same<T, float2>::value ||
+           metal::is_same<T, bf16_2>::value ||
+           metal::is_same<T, half2>::value;
+}
+    
+template <typename T>
+static METAL_FUNC constexpr const bool isT1Ptr() {
+    return metal::is_same<T, device      float*>::value ||
+           metal::is_same<T, threadgroup float*>::value ||
+           metal::is_same<T, thread      float*>::value ||
+           metal::is_same<T, device      bf16*>::value ||
+           metal::is_same<T, threadgroup bf16*>::value ||
+           metal::is_same<T, thread      bf16*>::value ||
+           metal::is_same<T, device      half*>::value ||
+           metal::is_same<T, threadgroup half*>::value ||
+           metal::is_same<T, thread      half*>::value;
+}
+template <typename T>
+static METAL_FUNC constexpr const bool isT2Ptr() {
+    return metal::is_same<T, device      float2*>::value ||
+           metal::is_same<T, threadgroup float2*>::value ||
+           metal::is_same<T, thread      float2*>::value ||
+           metal::is_same<T, device      bf16_2*>::value ||
+           metal::is_same<T, threadgroup bf16_2*>::value ||
+           metal::is_same<T, thread      bf16_2*>::value ||
+           metal::is_same<T, device      half2*>::value ||
+           metal::is_same<T, threadgroup half2*>::value ||
+           metal::is_same<T, thread      half2*>::value;
+}
+
+template <typename T>
+static METAL_FUNC constexpr const bool isTKType() { // good enough
+    return !isT1Type<T>() && !isT2Type<T>() && !isT1Ptr<T>() && !isT2Ptr<T>();
+}
+    
+} // namespace base_types
+} // namespace ducks
+    
+/**
+ * @namespace base_types
+ *
+ * @brief A namespace for Thundermittens basic data types.
+ */
+namespace base_types {
+/**
+ * @brief Provides compile-time constants for different types.
+ *
+ * @tparam T The type for which to provide constants.
+ */
+template<typename T> struct constants {
+    /**
+     * @brief Zero
+     * @return Constexpr zero with type T
+     */
+    static METAL_FUNC constexpr T zero()      { return T{0}; }
+    /**
+     * @brief One
+     * @return Constexpr one with type T
+     */
+    static METAL_FUNC constexpr T one()       { return T{1}; }
+    /**
+     * @brief Positive infinity. Particularly useful for initializing before a min op.
+     * @return Constexpr positive infinity with type T
+     */
+    static METAL_FUNC constexpr T pos_infty() { return T{INFINITY}; } // I'll find a better way at some point but this appears to work.
+    /**
+     * @brief Negative infinity. Particularly useful for initializing before a max op.
+     * @return Constexpr negative infinity with type T
+     */
+    static METAL_FUNC constexpr T neg_infty() { return T{-INFINITY}; }
+};
+template<> struct constants<float> {
+    static METAL_FUNC constexpr float zero()      { return 0.f; }
+    static METAL_FUNC constexpr float one()       { return 1.f; }
+    static METAL_FUNC constexpr float pos_infty() { return INFINITY; }
+    static METAL_FUNC constexpr float neg_infty() { return -INFINITY; }
+};
+template<> struct constants<float2> {
+    static METAL_FUNC constexpr float2 zero()      { return float2(0.f, 0.f); }
+    static METAL_FUNC constexpr float2 one()       { return float2(1.f, 1.f); }
+    static METAL_FUNC constexpr float2 pos_infty() { return float2(constants<float>::pos_infty(), constants<float>::pos_infty()); }
+    static METAL_FUNC constexpr float2 neg_infty() { return float2(constants<float>::neg_infty(), constants<float>::neg_infty()); }
+};
+template<> struct constants<bf16> {
+    static METAL_FUNC constexpr bf16 zero()      { return 0.bf; }
+    static METAL_FUNC constexpr bf16 one()       { return 1.bf; }
+    static METAL_FUNC constexpr bf16 pos_infty() { return HUGE_VALBF; }
+    static METAL_FUNC constexpr bf16 neg_infty() { return -HUGE_VALBF; }
+};
+template<> struct constants<bf16_2> {
+    static METAL_FUNC constexpr bf16_2 zero()      { return bf16_2(constants<bf16>::zero(),      constants<bf16>::zero());      }
+    static METAL_FUNC constexpr bf16_2 one()       { return bf16_2(constants<bf16>::one(),       constants<bf16>::one());       }
+    static METAL_FUNC constexpr bf16_2 pos_infty() { return bf16_2(constants<bf16>::pos_infty(), constants<bf16>::pos_infty()); }
+    static METAL_FUNC constexpr bf16_2 neg_infty() { return bf16_2(constants<bf16>::neg_infty(), constants<bf16>::neg_infty()); }
+};
+template<> struct constants<half> {
+    static METAL_FUNC constexpr half zero()      { return half(0.h); }
+    static METAL_FUNC constexpr half one()       { return half(1.h); }
+    static METAL_FUNC constexpr half pos_infty() { return HUGE_VALH; }
+    static METAL_FUNC constexpr half neg_infty() { return -HUGE_VALH; }
+};
+    
+template<> struct constants<half2> {
+    static METAL_FUNC constexpr half2 zero()      { return half2(constants<half>::zero(),      constants<half>::zero());      }
+    static METAL_FUNC constexpr half2 one()       { return half2(constants<half>::one(),       constants<half>::one());       }
+    static METAL_FUNC constexpr half2 pos_infty() { return half2(constants<half>::pos_infty(), constants<half>::pos_infty()); }
+    static METAL_FUNC constexpr half2 neg_infty() { return half2(constants<half>::neg_infty(), constants<half>::neg_infty()); }
+};
+
+    
+
+/**
+ * @brief Provides information about packing of elements for a given type.
+ *
+ * @tparam T The type for which to provide packing information.
+ */
+template<typename T> struct packing {
+//    /**
+//     * @brief The number of elements packed together.
+//     *
+//     * @return constexpr int representing number of elements within the type.
+//     */
+//    static METAL_FUNC constexpr int num() { return 1; }
+//    /**
+//     * @brief Packs a single T element twice (replicated) into its packed type.
+//     *
+//     * @param i[in] The element to pack.
+//     * @return The packed type.
+//     */
+//    static METAL_FUNC constexpr T pack(device const bf16 &i);
+//    static METAL_FUNC constexpr T pack(threadgroup const bf16 &i);
+//    static METAL_FUNC constexpr T pack(thread const bf16 &i);
+};
+
+#define PACK_FUNCTIONS(T1, T2) \
+    static METAL_FUNC constexpr T2 pack(device const T1 &i) { return T2{i, i};      } \
+    static METAL_FUNC constexpr T2 pack(threadgroup const T1 &i) { return T2{i, i}; } \
+    static METAL_FUNC constexpr T2 pack(thread const T1 &i) { return T2{i, i};      }
+
+template<> struct packing<bf16> {
+    static METAL_FUNC constexpr int num() { return 1; }
+    using unpacked_type = bf16;
+    using packed_type = bf16_2;
+    using packed_four = bf16_4;
+    PACK_FUNCTIONS(unpacked_type, packed_type)
+};
+template<> struct packing<half> {
+    static METAL_FUNC constexpr int num() { return 1; }
+    using unpacked_type = half;
+    using packed_type = half2;
+    using packed_four = half4;
+    PACK_FUNCTIONS(unpacked_type, packed_type)
+};
+template<> struct packing<float> {
+    static METAL_FUNC constexpr int num() { return 1; }
+    using unpacked_type = float;
+    using packed_type = float2;
+    using packed_four = float4;
+    
+    PACK_FUNCTIONS(unpacked_type, packed_type)
+};
+template<> struct packing<bf16_2> {
+    static METAL_FUNC constexpr int num() { return 2; }
+    using unpacked_type = bf16;
+    using packed_type = bf16_2;
+    using packed_four = bf16_4;
+    PACK_FUNCTIONS(unpacked_type, packed_type)
+};
+template<> struct packing<half2> {
+    static METAL_FUNC constexpr int num() { return 2; }
+    using unpacked_type = half;
+    using packed_type = half2;
+    using packed_four = half4;
+    PACK_FUNCTIONS(unpacked_type, packed_type)
+};
+template<> struct packing<float2> {
+    static METAL_FUNC constexpr int num() { return 2; }
+    using unpacked_type = float;
+    using packed_type = float2;
+    using packed_four = float4;
+    PACK_FUNCTIONS(unpacked_type, packed_type)
+};
+template<> struct packing<int2> {
+    static METAL_FUNC constexpr int num() { return 2; }
+};
+template<> struct packing<float4> {
+    static METAL_FUNC constexpr int num() { return 4; }
+};
+template<> struct packing<int4> {
+    static METAL_FUNC constexpr int num() { return 4; }
+};
+
+    
+/**
+ * @brief Provides templated functionality to convert between different types.
+ *
+ * @tparam T The target type for conversion.
+ * @tparam U The source type for conversion.
+ */
+template<typename T, typename U> struct convertor {
+    /**
+     * @brief Converts a value of type U to type T.
+     *
+     * @param u[in] The value of type U to convert.
+     * @return T The converted value of type T.
+     */
+    static METAL_FUNC T convert(device const U & u)      { return (T)u; }
+    static METAL_FUNC T convert(threadgroup const U & u) { return (T)u; }
+    static METAL_FUNC T convert(thread const U & u)      { return (T)u; }
+};
+    
+template<> struct convertor<float, bf16> {
+    // fptrunc float %_ to bfloat
+    static METAL_FUNC float convert(device const bf16 & u)      { return float(u);}
+    static METAL_FUNC float convert(threadgroup const bf16 & u) { return float(u);}
+    static METAL_FUNC float convert(thread const bf16 & u)      { return float(u);}
+};
+template<> struct convertor<bf16, float> {
+    // fpext bfloat %_ to float
+    static METAL_FUNC bf16 convert(device const float & u)      { return bf16(u); }
+    static METAL_FUNC bf16 convert(threadgroup const float & u) { return bf16(u); }
+    static METAL_FUNC bf16 convert(thread const float & u)      { return bf16(u); }
+};
+template<> struct convertor<float2, bf16_2> {
+    // tail call fast <2 x float> @air.convert.f.v2f32.f.v2bf16(<2 x bfloat> %_)
+    static METAL_FUNC float2 convert(device const bf16_2 & u)      { return float2(u); }
+    static METAL_FUNC float2 convert(threadgroup const bf16_2 & u) { return float2(u); }
+    static METAL_FUNC float2 convert(thread const bf16_2 & u)      { return float2(u); }
+};
+template<> struct convertor<bf16_2, float2> {
+    // tail call fast <2 x bfloat> @air.convert.f.v2bf16.f.v2f32(<2 x float> %_)
+    static METAL_FUNC bf16_2 convert(device const float2 & u)      { return bf16_2(u); }
+    static METAL_FUNC bf16_2 convert(threadgroup const float2 & u) { return bf16_2(u); }
+    static METAL_FUNC bf16_2 convert(thread const float2 & u)      { return bf16_2(u); }
+};
+    
+template<> struct convertor<float, half> {
+    // fptrunc float %_ to half
+    static METAL_FUNC float convert(device const half & u)      { return float(u); }
+    static METAL_FUNC float convert(threadgroup const half & u) { return float(u); }
+    static METAL_FUNC float convert(thread const half & u)      { return float(u); }
+};
+template<> struct convertor<half, float> {
+    //fpext half %_ to float
+    static METAL_FUNC half convert(device const float & u)      { return half(u); }
+    static METAL_FUNC half convert(threadgroup const float & u) { return half(u); }
+    static METAL_FUNC half convert(thread const float & u)      { return half(u); }
+};
+template<> struct convertor<float2, half2> {
+    // tail call fast <2 x float> @air.convert.f.v2f32.f.v2f16(<2 x half> %_)
+    static METAL_FUNC float2 convert(device const half2 & u)      { return float2(u); }
+    static METAL_FUNC float2 convert(threadgroup const half2 & u) { return float2(u); }
+    static METAL_FUNC float2 convert(thread const half2 & u)      { return float2(u); }
+};
+template<> struct convertor<half2, float2> {
+    // tail call fast <2 x half> @air.convert.f.v2f16.f.v2f32(<2 x float> %_)
+    static METAL_FUNC half2 convert(device const float2 & u)      { return half2(u); }
+    static METAL_FUNC half2 convert(threadgroup const float2 & u) { return half2(u); }
+    static METAL_FUNC half2 convert(thread const float2 & u)      { return half2(u); }
+};
+template<> struct convertor<bf16, half> {
+    static METAL_FUNC bf16 convert(device const half & u)      { return bf16(u); }
+    static METAL_FUNC bf16 convert(threadgroup const half & u) { return bf16(u); }
+    static METAL_FUNC bf16 convert(thread const half & u)      { return bf16(u); }
+};
+template<> struct convertor<half, bf16> {
+    static METAL_FUNC half convert(device const bf16 & u)      { return half(u); }
+    static METAL_FUNC half convert(threadgroup const bf16 & u) { return half(u); }
+    static METAL_FUNC half convert(thread const bf16 & u)      { return half(u); }
+};
+template<> struct convertor<bf16_2, half2> {
+    // tail call fast <2 x bfloat> @air.convert.f.v2bf16.f.v2f16(<2 x half> %_)
+    static METAL_FUNC bf16_2 convert(device const half2 & u)      { return bf16_2(u); }
+    static METAL_FUNC bf16_2 convert(threadgroup const half2 & u) { return bf16_2(u); }
+    static METAL_FUNC bf16_2 convert(thread const half2 & u)      { return bf16_2(u); }
+};
+template<> struct convertor<half2, bf16_2> {
+    // tail call fast <2 x half> @air.convert.f.v2f16.f.v2bf16(<2 x bfloat> %_)
+    static METAL_FUNC half2 convert(device const bf16_2 & u)      { return half2(u); }
+    static METAL_FUNC half2 convert(threadgroup const bf16_2 & u) { return half2(u); }
+    static METAL_FUNC half2 convert(thread const bf16_2 & u)      { return half2(u); }
+};
+    
+    
+    
+} // base_types
+
+} // mittens
@@ -0,0 +1,10 @@
+/**
+ * @file
+ * @brief A collection of common resources on which Thundermittens depends.
+ */
+
+
+#pragma once
+#include "base_types.metal"
+#include "base_ops.metal"
+#include "utils.metal"
@@ -0,0 +1,225 @@
+/**
+ * @file
+ * @brief General utilities for Thundermittens.
+ */
+#pragma once // not done
+/*
+ TODO:
+   shared allocator
+   max shared mem for other hardware
+ */
+
+#include <metal_stdlib>
+#include "base_types.metal"
+/**
+ * @namespace mittens
+ *
+ * @brief The main namespace of Thundermittens.
+ */
+namespace mittens {
+/**
+ * @namespace ore
+ *
+ * @brief The main namespace of Thundermittens Metal.
+ */
+
+/* ----------  GENERAL CONSTANTS FOR mittens  ---------- */
+
+/**
+ * @brief Tile dimension constant.
+ */
+constant constexpr const int TILE_DIM{8};
+constant constexpr const int TILE_ELEMENTS{TILE_DIM*TILE_DIM};
+constant constexpr const int SIMD_THREADS{32};
+    
+    
+#ifdef M2_PRO
+constant constexpr int MAX_SHARED_MEMORY = 32768;
+#else
+constant constexpr int MAX_SHARED_MEMORY = 32768;
+#endif
+/* ----------  TYPE HELPERS  ---------- */
+/**
+ * @namespace ducks
+ *
+ * @brief Thundermittens' namespace for template metaprogramming..
+ *
+ * This includes primarily dummy types and concept wrappers, along
+ * with a few additional utilities.
+ */
+namespace ducks {
+
+/**
+ * @brief A type representing an empty default for a template.
+ */
+struct default_type {};
+
+// This macro can't be done as a template, so it doesn't really have a location in mittens.
+#define typeof(A) typename std::remove_const<typename std::remove_reference<decltype(A)>::type>::type
+
+    
+}
+    
+/* ----------  SHUFFLE UTILS  ---------- */
+/**
+ * @brief Mask constant for all active threads in a warp.
+ */
+constant static constexpr uint32_t MASK_ALL = 0xFFFFFFFF;
+    
+template<typename T>
+static METAL_FUNC T shfl_sync(thread const T &f, const ushort laneid) {
+    return metal::simd_shuffle(f, laneid);
+}
+    
+template<>
+METAL_FUNC bfloat shfl_sync<bfloat>(thread const bf16 &f, const ushort laneid) {
+//    return as_type<bf16>(metal::simd_shuffle(*(thread half*)(&f), laneid));
+    float f_val = (float)f;
+    float shfl_val = metal::simd_shuffle(f_val, laneid);
+    return (bf16)shfl_val;
+}
+    
+template<>
+METAL_FUNC bfloat2 shfl_sync<bfloat2>(thread const bf16_2 &f, const ushort laneid) {
+//    return as_type<bf16_2>(metal::simd_shuffle(*(thread half2*)(&f), laneid));
+    float2 f_val = (float2)f;
+    float2 shfl_val = metal::simd_shuffle(f_val, laneid);
+    return (bf16_2)shfl_val;
+}
+    
+template<typename T>
+static METAL_FUNC T shfl_down_fill_sync(thread const T &f, thread const T& fill_data, const ushort laneid) {
+    return metal::simd_shuffle_and_fill_down(f, laneid, fill_data);
+}
+    
+template<>
+METAL_FUNC bfloat shfl_down_fill_sync<bfloat>(thread const bfloat &f, thread const bfloat &fill_data, const ushort laneid) {
+//    return as_type<bf16>(metal::simd_shuffle_and_fill_down(*(thread half*)(&f), *(thread half*)(&fill_data), laneid));
+    float f_val = (float)f;
+    float fill_data_f = (float)fill_data;
+    float shfl_val = metal::simd_shuffle_and_fill_down(f_val, fill_data_f, laneid);
+    return (bf16)shfl_val;
+}
+template<>
+METAL_FUNC bfloat2 shfl_down_fill_sync<bfloat2>(thread const bfloat2 &f, thread const bfloat2 &fill_data, const ushort laneid) {
+//    return as_type<bf16_2>(metal::simd_shuffle_and_fill_down(*(thread half2*)(&f), *(thread half2*)(&fill_data), laneid));
+    float2 f_val = (float2)f;
+    float2 fill_data_f = (float2)fill_data;
+    float2 shfl_val = metal::simd_shuffle_and_fill_down(f_val, fill_data_f, laneid);
+    return (bf16_2)shfl_val;
+}
+/**
+ * @brief Perform a shuffle down operation on a packed type synchronously across a warp.
+ * @tparam T The type of the value to be shuffled.
+ * @param mask[in] The mask of active threads.
+ * @param f[in] The value to be shuffled.
+ * @param delta[in] The number of positions to shuffle down.
+ * @return The result of the shuffle operation.
+ */
+template<typename T>
+static METAL_FUNC T shfl_down_sync(thread const T &f, int delta) {
+    return metal::simd_shuffle_rotate_down(f, delta);
+}
+
+template<>
+METAL_FUNC bfloat shfl_down_sync<bfloat>(thread const bf16 &f, int delta) {
+//    return base_types::convertor<bf16, float>::convert(metal::simd_shuffle_rotate_down(base_types::convertor<float, bf16>::convert(f), delta));
+//    return as_type<bf16>(metal::simd_shuffle_rotate_down(*(thread half*)(&f), delta));
+    float f_val = (float)f;
+    float shfl_val = metal::simd_shuffle_rotate_down(f_val, delta);
+    return (bf16)shfl_val;
+}
+
+template<>
+METAL_FUNC bfloat2 shfl_down_sync<bfloat2>(thread const bf16_2 &f, int delta) {
+//    return as_type<bf16_2>(metal::simd_shuffle_rotate_down(*(thread const half2*)(&f), delta));
+//    return base_types::convertor<bf16_2, float2>::convert(metal::simd_shuffle_rotate_down(base_types::convertor<float2, bf16_2>::convert(f), delta));
+    
+    float2 f_val = (float2)f;
+    float2 shfl_val = metal::simd_shuffle_rotate_down(f_val, delta);
+    return (bf16_2)shfl_val;
+//    return as_type<bf16_2>(metal::simd_shuffle_rotate_down(*(thread half2*)(&f), delta));
+}
+    
+    
+/* ----------  LOOP UNROLLING UTILS  ---------- */
+    
+namespace meta {
+template <int Start, int End, int Stride, bool = (Start < End)>
+struct unroll_i_in_range {
+    template<class F, typename... Args>
+    static METAL_FUNC void run(F f, Args... args) {
+        f(Start, args...);
+        unroll_i_in_range<Start + Stride, End, Stride>::run(f, args...);
+    }
+};
+
+template <int Start, int End, int Stride>
+struct unroll_i_in_range<Start, End, Stride, false> {
+    template<class F, typename... Args>
+    static METAL_FUNC void run(F, Args...) {
+    }
+};
+
+
+template <int Start, int End, int Stride, bool = (Start < End)>
+struct unroll_i_j_in_range_inner {
+    template<class F, typename... Args>
+    static METAL_FUNC void run(F f, int outerIndex, Args... args) {
+        f(outerIndex, Start, args...);
+        unroll_i_j_in_range_inner<Start + Stride, End, Stride>::run(f, outerIndex, args...);
+    }
+};
+
+template <int Start, int End, int Stride>
+struct unroll_i_j_in_range_inner<Start, End, Stride, false> {
+    template<class F, typename... Args>
+    static METAL_FUNC void run(F, int, Args...) {
+    }
+};
+
+template <int StartOuter, int EndOuter, int StrideOuter,
+          int StartInner, int EndInner, int StrideInner,
+          bool = (StartOuter < EndOuter)>
+struct unroll_i_j_in_range {
+    template<class F, typename... Args>
+    static METAL_FUNC void run(F f, Args... args) {
+        unroll_i_j_in_range_inner<StartInner, EndInner, StrideInner>::run(
+            f, StartOuter, args...
+        );
+        unroll_i_j_in_range<
+            StartOuter + StrideOuter, EndOuter, StrideOuter,
+            StartInner, EndInner, StrideInner
+        >::run(f, args...);
+    }
+};
+
+template <int StartOuter, int EndOuter, int StrideOuter,
+          int StartInner, int EndInner, int StrideInner>
+struct unroll_i_j_in_range<StartOuter, EndOuter, StrideOuter,
+                          StartInner, EndInner, StrideInner, false> {
+    template<class F, typename... Args>
+    static METAL_FUNC void run(F, Args...) {
+    }
+};
+    
+}
+
+
+template <int N>
+struct ReadVector {
+    float _[N];
+};
+
+/* ----------  SHARED MEMORY UTILS  ---------- */
+
+#define mittens_ALIGN_AS(n) alignas(n)
+#define mittens_DEFAULT_ALIGN mittens_ALIGN_AS(16)
+    
+/**
+ * @brief Dummy structure for alignment purposes. Needed for WGMMA and TMA calls.
+ */
+struct mittens_DEFAULT_ALIGN alignment_dummy { int dummy; };
+}
+
+
@@ -0,0 +1,24 @@
+/**
+ * @file
+ * @brief An aggregate header of all group (multi-warp) operations defined by Thundermittens
+ */
+
+#pragma once
+#include "../../common/common.metal"
+#include "../../types/types.metal"
+#include "../warp/warp.metal" // several group memory ops rely on underlying warp-scope ops
+namespace mittens {
+template<int N_WARPS>
+struct group {
+    constant static constexpr int GROUP_WARPS = N_WARPS; // This alias produces nice parallelism.
+    constant static constexpr int GROUP_THREADS = N_WARPS * mittens::SIMD_THREADS; // This alias produces nice parallelism.
+    static METAL_FUNC int simd_laneid(const unsigned threadIdx) { return threadIdx % mittens::SIMD_THREADS;         }
+    static METAL_FUNC int laneid     (const unsigned threadIdx) { return threadIdx % GROUP_THREADS;                      }
+    static METAL_FUNC int warpid     (const unsigned threadIdx) { return laneid(threadIdx) / mittens::SIMD_THREADS; }
+    static METAL_FUNC int groupid    (const unsigned threadIdx) { return threadIdx / GROUP_THREADS;                      }
+    #include "memory/memory.metal"
+    #include "shared/shared.metal"
+};
+    
+
+}
@@ -0,0 +1,2 @@
+#include "tile/tile.metal"
+#include "vec/vec.metal"
@@ -0,0 +1,132 @@
+
+/**
+ * @file
+ * @brief Functions for a group to collaboratively transfer data directly between global memory and registers and back.
+ */
+
+/**
+ * @brief Collaboratively loads data from a source array into row-major layout tiles.
+ *
+ * @tparam RT The row-major layout tile type.
+ * @tparam U The data type of the source array.
+ * @param dst[out] The destination tile to load data into.
+ * @param src[in] The source array to load data from. 
+ * @param row_stride[in] The stride in elements between rows in the source array.
+ */
+template<typename RT, typename GL>
+static METAL_FUNC typename metal::enable_if<ducks::is_row_register_tile<RT>() && ducks::is_global_layout<GL>(), void>::type
+load(thread RT &dst, thread const GL &_src, thread const coord &idx, const int threadIdx) {
+    using T  = typename RT::dtype;
+    using T2 = typename base_types::packing<T>::packed_type;
+    using U  = typename GL::dtype;
+    using U2 = typename base_types::packing<U>::packed_type;
+    const device U *src = (device U*)&_src.template get<RT>(idx);
+    const int row_stride = _src.row_stride();
+    
+    int warp_laneid = threadIdx % 32;
+    const int row_offset = dst.rows * warpid(threadIdx);
+    const short qid = warp_laneid / 4;
+    const short simd_y = row_offset + (qid & 4) + (warp_laneid / 2) % 4;
+    const short simd_x = (qid & 2) * 2 + (warp_laneid % 2) * 2;
+    #pragma clang loop unroll(full)
+    for(int i = 0; i < dst.height; i++) {
+        int row = simd_y + i * RT::tile_size;
+        #pragma clang loop unroll(full)
+        for(int j = 0; j < dst.width; j++) {
+            int col = simd_x + j * RT::tile_size;
+            T2 src2 = base_types::convertor<T2, U2>::convert(*((device U2*)(&src[row * row_stride + col])));
+            dst.tiles[i][j].data.thread_elements()[0] = src2[0];
+            dst.tiles[i][j].data.thread_elements()[1] = src2[1];
+        }
+    }
+}
+
+template<typename RT, typename GL>
+static METAL_FUNC typename metal::enable_if<ducks::is_col_register_tile<RT>() && ducks::is_global_layout<GL>(), void>::type
+load(thread RT &dst, thread const GL &_src, thread const coord &idx, const int threadIdx) {
+    using T  = typename RT::dtype;
+    using T2 = typename base_types::packing<T>::packed_type;
+    using U  = typename GL::dtype;
+    using U2 = typename base_types::packing<U>::packed_type;
+    const device U *src = (device U*)&_src.template get<RT>(idx);
+    const int row_stride = _src.row_stride();
+    
+    int warp_laneid = threadIdx % 32;
+    const int row_offset = dst.rows * warpid(threadIdx);
+    const short qid = warp_laneid / 4;
+    const short simd_y = row_offset + (qid & 2) * 2 + (warp_laneid % 2) * 2;;
+    const short simd_x = (qid & 4) + (warp_laneid / 2) % 4;
+    #pragma clang loop unroll(full)
+    for(int i = 0; i < dst.height; i++) {
+        int row = simd_y + i * RT::tile_size;
+        #pragma clang loop unroll(full)
+        for(int j = 0; j < dst.width; j++) {
+            int col = simd_x + j * RT::tile_size;
+            T2 src2 = base_types::convertor<T2, U2>::convert(*((device U2*)(&src[row * row_stride + col])));
+            dst.tiles[i][j].data.thread_elements()[0] = base_types::convertor<T, U>::convert(src[row * row_stride + col]);
+            dst.tiles[i][j].data.thread_elements()[1] = base_types::convertor<T, U>::convert(src[(row + 1) * row_stride + col]);
+        }
+    }
+}
+/**
+ * @brief Collaboratively stores data from register tiles to a destination array in global memory with a row-major layout.
+ *
+ * @tparam RT The register tile type with a row-major layout.
+ * @tparam U The data type of the destination array.
+ * @param[out] dst The destination array in global memory to store data into.
+ * @param[in] src The source register tile to store data from.
+ * @param row_stride[in] The stride in elements between rows in the destination array.
+ */
+template<typename RT, typename GL>
+static METAL_FUNC typename metal::enable_if<ducks::is_row_register_tile<RT>(), void>::type
+store(thread GL &_dst, thread const RT &src, thread const coord &idx, const int threadIdx) {
+    using T  = typename RT::dtype;
+    using T2 = typename base_types::packing<T>::packed_type;
+    using U  = typename GL::dtype;
+    using U2 = typename base_types::packing<U>::packed_type;
+    device U *dst = (device U*)&(_dst.template get<RT>(idx));
+    const int row_stride = _dst.row_stride();
+    int warp_laneid = simd_laneid(threadIdx);
+    const int row_offset = src.rows * warpid(threadIdx);
+    const short qid = warp_laneid / 4;
+    const short simd_y = row_offset + (qid & 4) + (warp_laneid / 2) % 4;
+    const short simd_x = (qid & 2) * 2 + (warp_laneid % 2) * 2;
+    #pragma clang loop unroll(full)
+    for(int i = 0; i < src.height; i++) {
+        int row = simd_y + i * RT::tile_size;
+        #pragma clang loop unroll(full)
+        for(int j = 0; j < src.width; j++) {
+            int col = simd_x + j * RT::tile_size;
+            U2 src2 = base_types::convertor<U2, T2>::convert(T2(src.tiles[i][j].data.thread_elements()[0], src.tiles[i][j].data.thread_elements()[1]));
+            *(device U2*)(&dst[row*row_stride + col]) = src2;
+        }
+    }
+}
+
+template<typename RT, typename GL>
+static METAL_FUNC typename metal::enable_if<ducks::is_col_register_tile<RT>(), void>::type
+store(thread GL &_dst, thread const RT &src, thread const coord &idx, const int threadIdx) {
+    using T  = typename RT::dtype;
+    using T2 = typename base_types::packing<T>::packed_type;
+    using U  = typename GL::dtype;
+    using U2 = typename base_types::packing<U>::packed_type;
+    device U *dst = (device U*)&(_dst.template get<RT>(idx));
+    const int row_stride = _dst.row_stride();
+    int warp_laneid = simd_laneid(threadIdx);
+    const int row_offset = src.rows * warpid(threadIdx);
+    const short qid = warp_laneid / 4;
+//    const short simd_y = row_offset + (qid & 4) + (warp_laneid / 2) % 4;
+//    const short simd_x = (qid & 2) * 2 + (warp_laneid % 2) * 2;
+    const short simd_y = row_offset + (qid & 2) * 2 + (warp_laneid % 2) * 2;
+    const short simd_x = (qid & 4) + (warp_laneid / 2) % 4;
+    #pragma clang loop unroll(full)
+    for(int i = 0; i < src.height; i++) {
+        int row = simd_y + i * RT::tile_size;
+        #pragma clang loop unroll(full)
+        for(int j = 0; j < src.width; j++) {
+            int col = simd_x + j * RT::tile_size;
+            dst[row*row_stride + col] = base_types::convertor<U, T>::convert(src.tiles[i][j].data.thread_elements()[0]);
+            dst[(row + 1) * row_stride + col] = base_types::convertor<U, T>::convert(src.tiles[i][j].data.thread_elements()[1]);
+        }
+    }
+}
@@ -0,0 +1,144 @@
+/**
+ * @file
+ * @brief Group (collaborative warp) ops for loading shared tiles from and storing to global memory.
+ */
+
+
+//template<typename ST, typename U>
+//static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+//load(int i,
+//     threadgroup ST *dst, device U* src,
+//     thread const int& group_laneid,
+//     thread const int& memcpy_per_row,
+//     thread const int& elem_per_memcpy,
+//     thread const int& row_stride)
+//{
+//    int idx = i * GROUP_THREADS + group_laneid;
+//    int row = idx / memcpy_per_row;
+//    int col = (idx*elem_per_memcpy) % ST::cols;
+//    if (row < ST::rows) {
+//        *(threadgroup float4*)(&(*dst)[{row, col}]) = *(device float4*)(&src[row*row_stride + col]);
+//    }
+//}
+
+
+template<typename ST, typename GL>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_global_layout<GL>(), void>::type
+load(threadgroup ST &dst, thread const GL &_src, thread const coord &idx, const int threadIdx) {
+    int group_laneid = threadIdx % GROUP_THREADS;
+    using T = typename ST::T;
+    using U = typename GL::dtype;
+    device U *src = (device U*)&_src.template get<ST>(idx);
+    const int row_stride = _src.row_stride();
+    using read_vector = ReadVector<1>;
+    // we can handle this many rows each time we run a memcpy_async
+    constexpr const int elem_per_memcpy = sizeof(read_vector)/sizeof(typename ST::dtype);
+    constexpr const int memcpy_per_row = ST::cols / elem_per_memcpy;
+    int total_calls = ((ST::height * ST::width + (N_WARPS-1))) * TILE_DIM*TILE_DIM / (N_WARPS*SIMD_THREADS*elem_per_memcpy); // round up
+    #pragma clang loop unroll(full)
+    for(int i = 0; i < total_calls; i++) {
+
+        int idx = i * GROUP_THREADS + group_laneid;
+        int row = idx / memcpy_per_row;
+        int col = (idx*elem_per_memcpy) % dst.cols;
+        if (row<dst.rows && col < dst.cols) {
+            *(threadgroup read_vector*)(&dst[{row, col}]) = *(device read_vector*)(&src[row*row_stride + col]);
+//            *(threadgroup float*)(&dst[{row, col}]) = 1.0f;
+        }
+    }
+//    dst[{0, 0}] = base_types::convertor<T, float>::convert(1.f);
+//    dst[{0, 0}] = total_calls;
+//    meta::unroll_i_in_range<0, total_calls, 1>::run(load<ST, typename GL::dtype>, &dst, src, group_laneid, memcpy_per_row, elem_per_memcpy, row_stride);
+}
+
+
+//template<typename ST, typename GL>
+//static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_global_layout<GL>(), void>::type
+//load(threadgroup ST &dst, thread const GL &_src, thread const coord &idx, const int threadIdx) {
+//    int group_laneid = threadIdx % GROUP_THREADS;
+//    int groupid = threadIdx / GROUP_THREADS;
+//    int laneid = threadIdx % SIMD_THREADS;
+//    
+//    using U = typename GL::dtype;
+//    device U *src = (device U*)&_src.template get<ST>(idx);
+//    const int row_stride = _src.row_stride();
+//    
+//    int elem_per_memcpy = sizeof(float)/sizeof(typename ST::dtype);
+//    int memcpy_per_row = ST::cols / elem_per_memcpy;
+//    int total_calls = ((ST::height * ST::width + (N_WARPS-1))) * TILE_DIM*TILE_DIM / (N_WARPS*SIMD_THREADS*elem_per_memcpy); // round up
+//    /*
+//     1x16 or 8 x 128
+//     */
+//    int offset = ST::num_elements / (GROUP_WARPS);
+////    int offset = group_laneid
+//    #pragma clang loop unroll(full)
+//    for(int i = 0; i < total_calls; i++) {
+//        int idx = i * SIMD_THREADS + laneid;
+////        int idx = i * () + group_laneid;
+//        int row = idx / memcpy_per_row;
+//        int col = (idx*elem_per_memcpy) % dst.cols;
+//        if (row<dst.rows) {
+//            *(threadgroup float*)(&dst[{row, col}]) = *(device float*)(&src[row*row_stride + col]);
+//        }
+//    }
+//}
+//
+//template<typename ST, typename GL>
+//static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_global_layout<GL>(), void>::type
+//load(threadgroup ST &dst, thread const GL &_src, thread const coord &idx, const int threadIdx) {
+//    int warp_id = threadIdx / SIMD_THREADS;
+//    int lane_id = threadIdx % SIMD_THREADS;
+////    int N_WARPS = /* number of warps in your group */;
+//    
+//    using U = typename GL::dtype;
+//    device U *src = (device U*)&_src.template get<ST>(idx);
+//    const int row_stride = _src.row_stride();
+//    
+//    int elem_per_memcpy = sizeof(float)/sizeof(typename ST::dtype);
+//    int memcpy_per_row = ST::cols / elem_per_memcpy;
+//    int total_memcpy_elems = (ST::height * ST::cols) / elem_per_memcpy;
+//    int elems_per_warp = (total_memcpy_elems + N_WARPS - 1) / N_WARPS;  // Ceiling division
+//    
+//    int start_idx = warp_id * elems_per_warp;
+//    int end_idx = metal::min(start_idx + elems_per_warp, total_memcpy_elems);
+//    
+//    #pragma clang loop unroll(full)
+//    for (int idx = start_idx + lane_id; idx < end_idx; idx += SIMD_THREADS) {
+//        int row = idx / memcpy_per_row;
+//        int col = (idx % memcpy_per_row) * elem_per_memcpy;
+//        if (row < ST::height) {
+//            *(threadgroup float*)(&dst[{row, col}]) = *(device float*)(&src[row * row_stride + col]);
+//        }
+//    }
+//}
+
+template<typename ST, typename GL>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_global_layout<GL>(), void>::type
+store(thread const GL &_dst, threadgroup const ST &src, thread const coord &idx, const int threadIdx) {
+    int group_laneid = threadIdx % GROUP_THREADS;
+    using U = typename GL::dtype;
+    device U *dst = (device U*)&_dst.template get<ST>(idx);
+    const int row_stride = _dst.row_stride();
+    using read_vector = ReadVector<1>;
+    // we can handle this many rows each time we run a memcpy_async
+    int elem_per_memcpy = sizeof(read_vector)/sizeof(typename ST::dtype); // float/float -> 1
+    int memcpy_per_row = ST::cols / elem_per_memcpy; // 240 memcpy per row
+    int total_calls = ((src.height * src.width + (N_WARPS-1))) * TILE_DIM*TILE_DIM / (N_WARPS*SIMD_THREADS*elem_per_memcpy); // round up
+    
+    #pragma clang loop unroll(full)
+    for(int i = 0; i < total_calls; i++) {
+
+        int idx = i * GROUP_THREADS + group_laneid;
+        
+        int row = idx / memcpy_per_row;
+        int col = (idx*elem_per_memcpy) % src.cols;
+        if (row<src.rows && col < src.cols) {
+            *(device read_vector*)(&dst[row*row_stride + col]) = *(threadgroup read_vector*)(&src[{row, col}]);
+//            *(device float*)(&dst[row*row_stride + col]) = 1.f;
+        }
+    }
+//    dst[0] = src[{0,0}];
+//    dst[0] = total_calls;
+//    dst[0] = base_types::convertor<U, float>::convert(1);
+}
+
@@ -0,0 +1,152 @@
+/**
+ * @file
+ * @brief Functions for a warpgroup to collaboratively transfer data directly between shared memory and registers and back.
+ */
+
+/**
+ * @brief Collaboratively load data from a shared tile into register tiles split across a warpgroup.
+ *
+ * @tparam RT The register tile type
+ * @tparam ST The shared tile type
+ * @param dst[out] The destination register tile.
+ * @param src[in]  The source shared tile.
+ */
+template<typename RT, typename ST>
+METAL_FUNC static typename metal::enable_if<ducks::is_row_register_tile<RT>() && ducks::is_shared_tile<ST>(), void>::type
+load(thread RT &dst, threadgroup const ST &src, const int threadIdx) {
+    constexpr int height = ST::height;
+    constexpr int warp_height = RT::height;
+    static_assert(height%N_WARPS == 0, "Group load / store requires tile height to be a multiple of N_WARPS.");
+    static_assert(height%warp_height == 0, "Group load / store requires tile height to be a multiple of the RT height.");
+    static_assert(warp_height * N_WARPS == height, "RT height * N_WARPS must = ST height");
+    static_assert(ST::width==RT::width, "Group load / store requires tile widths to match.");
+    using T  = typename RT::dtype;
+    using T2 = typename base_types::packing<T>::packed_type;
+    using U  = typename ST::dtype;
+    using U2 = typename base_types::packing<U>::packed_type;
+    
+    int warp_laneid = simd_laneid(threadIdx);
+    const int row_offset = RT::rows * warpid(threadIdx);
+    const short qid = warp_laneid / 4;
+    const short simd_y = row_offset + (qid & 4) + (warp_laneid / 2) % 4;
+    const short simd_x = (qid & 2) * 2 + (warp_laneid % 2) * 2;
+    #pragma clang loop unroll(full)
+    for(int i = 0; i < dst.height; i++) {
+        int row = simd_y + i * mittens::TILE_DIM;
+        #pragma clang loop unroll(full)
+        for(int j = 0; j < dst.width; j++) {
+            int col = simd_x + j * mittens::TILE_DIM;
+            T2 src2 = base_types::convertor<T2, U2>::convert(*((threadgroup U2*)(&src[{row, col}])));
+            dst.tiles[i][j].data.thread_elements()[0] = src2[0];
+            dst.tiles[i][j].data.thread_elements()[1] = src2[1];
+        }
+    } 
+}
+
+template<typename RT, typename ST>
+METAL_FUNC static typename metal::enable_if<ducks::is_col_register_tile<RT>() && ducks::is_shared_tile<ST>(), void>::type
+load(thread RT &dst, threadgroup const ST &src, const int threadIdx) {
+    constexpr int height = ST::height;
+    constexpr int warp_height = RT::height;
+    static_assert(height%N_WARPS == 0, "Group load / store requires tile height to be a multiple of N_WARPS.");
+    static_assert(height%warp_height == 0, "Group load / store requires tile height to be a multiple of the RT height.");
+    static_assert(warp_height * N_WARPS == height, "RT height * N_WARPS must = ST height");
+    static_assert(ST::width==RT::width, "Group load / store requires tile widths to match.");
+    using T  = typename RT::dtype;
+    using T2 = typename base_types::packing<T>::packed_type;
+    using U  = typename ST::dtype;
+    using U2 = typename base_types::packing<U>::packed_type;
+    
+    int warp_laneid = simd_laneid(threadIdx);
+    const int row_offset = RT::rows * warpid(threadIdx);
+    const short qid = warp_laneid / 4;
+    const short simd_y = row_offset + (qid & 2) * 2 + (warp_laneid % 2) * 2;
+    const short simd_x = (qid & 4) + (warp_laneid / 2) % 4;
+    #pragma clang loop unroll(full)
+    for(int i = 0; i < dst.height; i++) {
+        #pragma clang loop unroll(full)
+        for(int j = 0; j < dst.width; j++) {
+            int row = simd_y + i * mittens::TILE_DIM;
+            int col = simd_x + j * mittens::TILE_DIM;
+            dst.tiles[i][j].data.thread_elements()[0] = base_types::convertor<T, U>::convert(src[{row + 0, col}]);
+            dst.tiles[i][j].data.thread_elements()[1] = base_types::convertor<T, U>::convert(src[{row + 1, col}]);
+        }
+    }
+}
+
+/**
+ * @brief Collaboratively store data into a shared tile from register tiles split across a warpgroup.
+ *
+ * @tparam RT The register tile type
+ * @tparam ST The shared tile type
+ * @param dst[out] The destination shared tile.
+ * @param src[in]  The source register tile.
+ */
+template<typename ST, typename RT>
+METAL_FUNC static typename metal::enable_if<ducks::is_row_register_tile<RT>() && ducks::is_shared_tile<ST>(), void>::type
+store(threadgroup ST &dst, thread const RT &src, const int threadIdx) {
+    constexpr int height = ST::height;
+    constexpr int warp_height = RT::height;
+    static_assert(height%N_WARPS == 0, "Group load / store requires tile height to be a multiple of N_WARPS.");
+    static_assert(height%warp_height == 0, "Group load / store requires tile height to be a multiple of the RT height.");
+    static_assert(warp_height * N_WARPS == height, "RT height * N_WARPS must = ST height");
+    static_assert(ST::width==RT::width, "Group load / store requires tile widths to match.");
+    using T  = typename RT::dtype;
+    using T2 = typename base_types::packing<T>::packed_type;
+    using U  = typename ST::dtype;
+    using U2 = typename base_types::packing<U>::packed_type;
+    int warp_laneid = simd_laneid(threadIdx);
+    const int row_offset = RT::rows * warpid(threadIdx);
+    const short qid = warp_laneid / 4;
+    const short simd_y = row_offset + (qid & 4) + (warp_laneid / 2) % 4;
+    const short simd_x = (qid & 2) * 2 + (warp_laneid % 2) * 2;
+    #pragma clang loop unroll(full)
+    for(int i = 0; i < RT::height; i++) {
+        int row = simd_y + i * mittens::TILE_DIM;
+        #pragma clang loop unroll(full)
+        for(int j = 0; j < RT::width; j++) {
+            int col = simd_x + j * mittens::TILE_DIM;
+            U2 src2 = base_types::convertor<U2, T2>::convert(T2(src.tiles[i][j].data.thread_elements()[0],
+                                                                src.tiles[i][j].data.thread_elements()[1]));
+            *(threadgroup U2*)(&dst[{row, col}]) = src2;
+        }
+    }
+}
+
+
+template<typename ST, typename RT>
+METAL_FUNC static typename metal::enable_if<ducks::is_col_register_tile<RT>() && ducks::is_shared_tile<ST>(), void>::type
+store(threadgroup ST &dst, thread const RT &src, const int threadIdx) {
+    constexpr int height = ST::height;
+    constexpr int warp_height = RT::height;
+    static_assert(height%N_WARPS == 0, "Group load / store requires tile height to be a multiple of N_WARPS.");
+    static_assert(height%warp_height == 0, "Group load / store requires tile height to be a multiple of the RT height.");
+    static_assert(warp_height * N_WARPS == height, "RT height * N_WARPS must = ST height");
+    static_assert(ST::width==RT::width, "Group load / store requires tile widths to match.");
+    using T  = typename RT::dtype;
+    using T2 = typename base_types::packing<T>::packed_type;
+    using U  = typename ST::dtype;
+    using U2 = typename base_types::packing<U>::packed_type;
+    int warp_laneid = simd_laneid(threadIdx);
+    const int row_offset = RT::rows * warpid(threadIdx);
+    const short qid = warp_laneid / 4;
+//    const short simd_y = row_offset + (qid & 4) + (warp_laneid / 2) % 4;
+//    const short simd_x = (qid & 2) * 2 + (warp_laneid % 2) * 2;
+    const short simd_y = row_offset + (qid & 2) * 2 + (warp_laneid % 2) * 2;
+    const short simd_x = (qid & 4) + (warp_laneid / 2) % 4;
+    #pragma clang loop unroll(full)
+    for(int i = 0; i < RT::height; i++) {
+        
+        #pragma clang loop unroll(full)
+        for(int j = 0; j < RT::width; j++) {
+            int row = simd_y + i * mittens::TILE_DIM;
+            int col = simd_x + j * mittens::TILE_DIM;
+//            U2 src2 = base_types::convertor<U2, T2>::convert(T2(src.tiles[i][j].data.thread_elements()[0],
+//                                                                src.tiles[i][j].data.thread_elements()[1]));
+//            *(threadgroup U2*)(&dst[{row, col}]) = src2;
+            
+            dst[{row + 0, col}] = base_types::convertor<U, T>::convert(src.tiles[i][j].data.thread_elements()[0]);
+            dst[{row + 1, col}] = base_types::convertor<U, T>::convert(src.tiles[i][j].data.thread_elements()[1]);
+        }
+    }
+}
@@ -0,0 +1,8 @@
+/**
+ * @file
+ * @brief An aggregate header of group memory operations on tiles.
+ */
+
+#include "shared_to_register.metal"
+#include "global_to_register.metal"
+#include "global_to_shared.metal"
@@ -0,0 +1,47 @@
+
+/**
+ * @file
+ * @brief Functions for a warpgroup to collaboratively transfer  data directly between global memory and registers and back.
+ */
+
+/**
+ * @brief Collaboratively loads data into register vectors from a source array in global memory.
+ *
+ * @tparam RV The register vector type.
+ * @tparam U The data type of the source array.
+ * @param[out] dst The destination register vector to load data into.
+ * @param[in] src The source array in global memory to load data from.
+ */
+template<typename RV, typename GL>
+METAL_FUNC static typename metal::enable_if<ducks::is_register_vector<RV>(), void>::type
+load(thread RV &dst, thread const GL &_src, thread coord idx, const int threadIdx) {
+    using T  = typename RV::dtype;
+    using U  = typename GL::dtype;
+    using U2 = typename base_types::packing<U>::packed_type;
+    using T2 = typename base_types::packing<T>::packed_type;
+    
+    idx.c += warpid(threadIdx);
+    // Call warp level store
+    ::mittens::load(dst, _src, idx, simd_laneid(threadIdx));
+}
+
+/**
+ * @brief Collaboratively stores data from register vectors to a destination array in global memory.
+ *
+ * @tparam RV The register vector type.
+ * @tparam U The data type of the destination array.
+ * @param[out] dst The destination array in global memory to store data into.
+ * @param[in] src The source register vector to store data from.
+ */
+template<typename RV, typename GL>
+METAL_FUNC static typename metal::enable_if<ducks::is_register_vector<RV>(), void>::type
+store(thread GL &_dst, thread const RV &src, thread coord idx, const int threadIdx) {
+    using T  = typename RV::dtype;
+//    using U2 = typename base_types::packing<U>::packed_type;
+    using T2 = typename base_types::packing<T>::packed_type;
+    
+    idx.c += warpid(threadIdx);
+
+    // Call warp level store
+    ::mittens::store(_dst, src, idx, simd_laneid(threadIdx));
+}
@@ -0,0 +1,59 @@
+/**
+ * @file
+ * @brief Group (collaborative warp) ops for loading shared vectors from and storing to global memory.
+ */
+
+/**
+ * @brief Loads data from global memory into shared memory vector.
+ *
+ * This function loads data from a global memory location pointed to by `src` into a shared memory vector `dst`.
+ * It calculates the number of elements that can be transferred in one operation based on the size ratio of `float4` to the data type of `SV`.
+ * The function ensures coalesced memory access and efficient use of bandwidth by dividing the work among threads in a warp.
+ *
+ * @tparam SV Shared vector type, must satisfy ducks::sv::all concept.
+ * @param dst Reference to the shared vector where the data will be loaded.
+ * @param src Pointer to the global memory location from where the data will be loaded.
+ */
+template<typename SV, typename GL>
+METAL_FUNC static typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+load(threadgroup SV &dst, thread const GL &_src, thread const coord &idx, const int threadIdx) {
+    using U = typename GL::dtype;
+    using read_vector = ReadVector<1>;
+    constexpr int elem_per_transfer = sizeof(read_vector) / sizeof(typename SV::dtype);
+    constexpr int total_calls = SV::length / elem_per_transfer; // guaranteed to divide
+    device U *src = (device U*)&_src.template get<SV>(idx);
+    
+    #pragma clang loop unroll(full)
+    for(int i = laneid(threadIdx); i < total_calls; i+=GROUP_THREADS) {
+        if(i * elem_per_transfer < dst.length)
+            *(threadgroup read_vector*)&dst[i*elem_per_transfer] = *(device read_vector*)&src[i*elem_per_transfer];
+    }
+}
+ 
+/**
+ * @brief Stores data from a shared memory vector to global memory.
+ *
+ * This function stores data from a shared memory vector `src` to a global memory location pointed to by `dst`.
+ * Similar to the load function, it calculates the number of elements that can be transferred in one operation based on the size ratio of `float4` to the data type of `SV`.
+ * The function ensures coalesced memory access and efficient use of bandwidth by dividing the work among threads in a warp.
+ *
+ * @tparam SV Shared vector type, must satisfy ducks::sv::all concept.
+ * @param dst Pointer to the global memory location where the data will be stored.
+ * @param src Reference to the shared vector from where the data will be stored.
+ */
+template<typename SV, typename GL>
+METAL_FUNC static typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+store(thread const GL &_dst, threadgroup const SV &src, thread const coord &idx, const int threadIdx) {
+    using read_vector = ReadVector<1>;
+    using U = typename GL::dtype;
+    constexpr int elem_per_transfer = sizeof(read_vector) / sizeof(typename SV::dtype);
+    constexpr int total_calls = SV::length / elem_per_transfer; // guaranteed to divide
+    device U *dst = (device U*)&_dst.template get<SV>(idx);
+    
+    metal::simdgroup_barrier(metal::mem_flags::mem_none);
+    #pragma clang loop unroll(full)
+    for(int i = laneid(threadIdx); i < total_calls; i+= GROUP_THREADS) {
+        if(i * elem_per_transfer < src.length)
+            *(device read_vector*)&dst[i*elem_per_transfer] = *(threadgroup read_vector*)&src[i*elem_per_transfer]; // lmao it's identical
+    }
+}
@@ -0,0 +1,60 @@
+/**
+ * @file
+ * @brief Functions for a group to collaboratively transfer data directly between shared memory and registers and back.
+ */
+
+/**
+ * @brief Collaboratively load data from a shared vector into register vectors split across a warpgroup.
+ *
+ * @tparam RV The register vector type
+ * @tparam SV The shared vector type
+ * @param dst[out] The destination register vector.
+ * @param src[in]  The source shared vector.
+ */
+template<typename RV, typename SV>
+METAL_FUNC static typename metal::enable_if<ducks::is_register_vector<RV>() && ducks::is_shared_vector<SV>(), void>::type
+load(thread RV &dst, threadgroup const SV &_src, const int threadIdx) {
+    using T  = typename RV::dtype;
+    using U  = typename SV::dtype;
+    using U2 = typename base_types::packing<U>::packed_type;
+    using T2 = typename base_types::packing<T>::packed_type;
+
+    static_assert(SV::length == RV::length*N_WARPS, "rv and sv dimensions do not match");// confirm size correct
+//    threadgroup typename SV::template subvec<typename SV::dtype, RV::outer_dim> &src = subvec_inplace<RV::outer_dim, SV>(_src, warpid(threadIdx));
+    //    threadgroup subvec &src = subvec_inplace<RV::outer_dim, SV>(_src, warpid(threadIdx));
+    unsigned warpId = warpid(threadIdx);
+    using subvec = typename SV::template subvec<RV::length>;
+
+    threadgroup subvec& src = *(threadgroup subvec*)(&_src[warpId *RV::length]);
+    
+    ::mittens::load<RV, subvec>(dst, src, simd_laneid(threadIdx)); // warp-level
+}
+
+/**
+ * @brief Collaboratively store data into a shared vector from register vectors split across a warpgroup.
+ *
+ * @tparam RV The register vector type
+ * @tparam SV The shared vector type
+ * @param dst[out] The destination shared vector.
+ * @param src[in]  The source register vector.
+ */
+template<typename SV, typename RV>
+METAL_FUNC static typename metal::enable_if<ducks::is_register_vector<RV>() && ducks::is_shared_vector<SV>(), void>::type
+store(threadgroup SV &_dst, thread const RV &src, const int threadIdx) {
+    using T  = typename RV::dtype;
+    using U  = typename SV::dtype;
+    using T2 = typename base_types::packing<T>::packed_type;
+    using U2 = typename base_types::packing<U>::packed_type;
+    
+
+    static_assert(SV::length == RV::length*N_WARPS, "rv and sv dimensions do not match");// confirm size correct
+    
+//    threadgroup typename SV::template subvec<typename SV::dtype, RV::outer_dim> &dst = subvec_inplace<RV::outer_dim, SV>(_dst, warpid(threadIdx));
+//    ::mittens::store<threadgroup typename SV::template subvec<typename SV::dtype, RV::outer_dim>, RV>(dst, src, simd_laneid(threadIdx)); // warp-level
+    
+    unsigned warpId = warpid(threadIdx);
+    using subvec = typename SV::template subvec<RV::length>;
+    threadgroup subvec& dst = *(threadgroup subvec*)(&_dst[warpId * RV::length]);
+    
+    ::mittens::store(dst, src, simd_laneid(threadIdx)); // warp-level
+}
@@ -0,0 +1,8 @@
+/**
+ * @file
+ * @brief An aggregate header of group memory operations on vectors.
+ */
+
+#include "shared_to_register.metal"
+#include "global_to_register.metal"
+#include "global_to_shared.metal"
@@ -0,0 +1,3 @@
+
+#include "tile/tile.metal"
+#include "vec/vec.metal"
@@ -0,0 +1,27 @@
+/**
+ * @file
+ * @brief Group conversions between different shared memory tile types.
+ */
+
+/* ----------  COPIES  ---------- */
+
+/**
+ * @brief Copies data from one shared memory tile to another, potentially with different data types and layouts.
+ *
+ * @tparam T The data type of the destination tile.
+ * @tparam U The data type of the source tile.
+ * @tparam _height The height of the tile.
+ * @tparam _width The width of the tile.
+ * @tparam L1 The layout of the destination tile.
+ * @tparam L2 The layout of the source tile.
+ * @param[out] dst The destination tile.
+ * @param[in] src The source tile.
+ */
+template<typename T, typename U, int _height, int _width>
+static METAL_FUNC void copy(threadgroup st<T, _height, _width> &dst, threadgroup const st<U, _height, _width> &src, const int threadIdx) {
+    #pragma clang loop unroll(full)
+    for(int i = laneid(threadIdx); i < dst.num_elements; i+=GROUP_THREADS) {
+        int row = i/dst.cols, col = i%dst.cols;
+        dst[{row, col}] = base_types::convertor<T, U>::convert(src[{row, col}]);
+    }
+}
@@ -0,0 +1,475 @@
+/**  
+ * @file
+ * @brief Group maps on shared tiles.
+ */
+
+/**
+ * @brief Performs a uniform unary operation on a tile.
+ *
+ * This function applies a given unary operation to each element of the source tile and stores the result in the destination tile.
+ * The operation is applied independently to each element, without considering its position or the values of neighboring elements.
+ *
+ * @tparam op The unary operation to be applied. Must be specialized to support operation on the data type of T.
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] src The source tile to which the unary operation is applied.
+ */
+template<typename op, typename ST> // T2, w, h can be inferred from dst as long as op is specialized
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ unary_map(threadgroup ST &dst, threadgroup const ST &src, const int threadIdx) {
+    #pragma clang loop unroll(full)
+    for(int i = laneid(threadIdx); i < dst.num_elements; i += GROUP_THREADS) {
+        dst.data[i] = op::template op<typename ST::dtype>(src.data[i]);
+    }
+}
+
+/**
+ * @brief Performs a uniform binary operation on a tile with a scalar parameter.
+ *
+ * This function applies a given binary operation to each element of the source tile and a scalar parameter, then stores the result in the destination tile.
+ * The operation is applied independently to each element, treating the scalar parameter as the second operand for each operation.
+ *
+ * @tparam op The binary operation to be applied. Must be specialized to support operation on the data type of T and the scalar parameter.
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] src The source tile to which the binary operation is applied.
+ * @param[in] param The scalar parameter to be used as the second operand in the binary operation.
+ */
+template<typename op, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ bin_map(threadgroup ST &dst, threadgroup const ST &src, thread const typename ST::dtype &param, const int threadIdx) {
+    #pragma clang loop unroll(full)
+    for(int i = laneid(threadIdx); i < dst.num_elements; i += GROUP_THREADS) {
+        dst.data[i] = op::template op<typename ST::dtype>(src.data[i], param);
+    }
+}
+
+/**
+ * @brief Performs a uniform binary operation on two tiles.
+ *
+ * This function applies a given binary operation to corresponding elements of two source tiles and stores the result in the destination tile.
+ * The operation is applied independently to each pair of elements, without considering their positions or the values of neighboring elements.
+ *
+ * @tparam op The binary operation to be applied. Must be specialized to support operation on the data type of T.
+ * @tparam T The type of the tiles. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] lhs The first source tile to which the binary operation is applied.
+ * @param[in] rhs The second source tile to which the binary operation is applied.
+ */
+template<typename op, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ bin_map(threadgroup ST &dst, threadgroup const ST &lhs, threadgroup const ST &rhs, const int threadIdx) {
+    #pragma clang loop unroll(full)
+    for(int i = laneid(threadIdx); i < dst.num_elements; i += GROUP_THREADS) {
+        dst.data[i] = op::template op<typename ST::dtype>(lhs.data[i], rhs.data[i]);
+    }
+}
+
+/**
+ * @brief Performs a row-wise binary operation on a tile with a vector.
+ *
+ * This function applies a given binary operation to each row of the source tile and the corresponding element of the source vector,
+ * then stores the result in the destination tile. The operation is applied independently to each row, using the vector element as
+ * the second operand for each element in the row.
+ *
+ * @tparam op The binary operation to be applied. Must be specialized to support operation on the data type of T and the vector elements.
+ * @tparam T The type of the tiles. Must satisfy the `ducks::st::all` concept.
+ * @tparam V The type of the vector. Must have the same data type as T.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] src The source tile to which the binary operation is applied.
+ * @param[in] vec The source vector containing the second operand for each row operation.
+ */
+template<typename op, typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+row_map(threadgroup ST &dst, threadgroup const ST &src, threadgroup const SV &vec, const int threadIdx) {
+    static_assert(metal::is_same<typename ST::dtype, typename SV::dtype>::value, "Tile and vector must have the same data type");
+    static_assert(SV::length == ST::rows, "Vector length must match the number of rows in the tile");
+    #pragma clang loop unroll(full)
+    for(int i = laneid(threadIdx); i < dst.num_elements; i += GROUP_THREADS) {
+        int row = i/dst.cols, col = i%dst.cols;
+        dst[{row, col}] = op::template op<typename ST::dtype>(src[{row, col}], vec[row]);
+    }
+}
+
+/**
+ * @brief Performs a column-wise binary operation on a tile with a vector.
+ *
+ * This function applies a given binary operation to each column of the source tile and the corresponding element of the source vector,
+ * then stores the result in the destination tile. The operation is applied independently to each column, using the vector element as
+ * the second operand for each element in the column.
+ *
+ * @tparam op The binary operation to be applied. Must be specialized to support operation on the data type of T and the vector elements.
+ * @tparam T The type of the tiles. Must satisfy the `ducks::st::all` concept.
+ * @tparam V The type of the vector. Must have the same data type as T.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] src The source tile to which the binary operation is applied.
+ * @param[in] vec The source vector containing the second operand for each column operation.
+ */
+template<typename op, typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ col_map(threadgroup ST &dst, threadgroup const ST &src, threadgroup const SV &vec, const int threadIdx) {
+    static_assert(metal::is_same<typename ST::dtype, typename SV::dtype>::value, "Tile and vector must have the same data type");
+    static_assert(SV::length == ST::cols, "Vector length must match the number of columns in the tile");
+    #pragma clang loop unroll(full)
+    for(int i = laneid(threadIdx); i < dst.num_elements; i += GROUP_THREADS) {
+        int row = i/dst.cols, col = i%dst.cols;
+        dst[{row, col}] = op::template op<typename ST::dtype>(src[{row, col}], vec[col]);
+    }
+}
+
+
+/* ----------  WRAPPERS FOR PRETTINESS  ---------- */
+
+// All of the annoying qualifiers *should* be automatically inferred during compile-time.
+// So, syntax should just be mittens::add_row(tile, colvec);
+
+// const maps
+/**
+ * @brief Sets all elements of the destination tile to zero.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile.
+ */
+template<typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ zero(threadgroup ST &dst, const int threadIdx) {
+    unary_map<base_ops::zero, ST>(dst, dst, threadIdx);
+}
+/**
+ * @brief Sets all elements of the destination tile to one.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile.
+ */
+template<typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ one(threadgroup ST &dst, const int threadIdx) {
+    unary_map<base_ops::one, ST>(dst, dst, threadIdx);
+}
+/**
+ * @brief Sets all elements of the destination tile to positive infinity.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile.
+ */
+template<typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ pos_infty(threadgroup ST &dst, const int threadIdx) {
+    unary_map<base_ops::pos_infty, ST>(dst, dst, threadIdx);
+}
+/**
+ * @brief Sets all elements of the destination tile to negative infinity.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile.
+ */
+template<typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ neg_infty(threadgroup ST &dst, const int threadIdx) {
+    unary_map<base_ops::neg_infty, ST>(dst, dst, threadIdx);
+}
+
+// unary maps
+/**
+ * @brief Applies the exponential function to each element of the source tile and stores the result in the destination tile.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] src The source tile to which the exponential function is applied.
+ */
+template<typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ exp(threadgroup ST &dst, threadgroup const ST &src, const int threadIdx) {
+    unary_map<base_ops::exp, ST>(dst, src, threadIdx);
+}
+/**
+ * @brief Applies the exponential function to each element of the source tile and stores the result in the destination tile, in base 2.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] src The source tile to which the exponential function is applied.
+ */
+template<typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ exp2(threadgroup ST &dst, threadgroup const ST &src, const int threadIdx) {
+    unary_map<base_ops::exp2, ST>(dst, src, threadIdx);
+}
+/**
+ * @brief Applies the natural logarithm function to each element of the source tile and stores the result in the destination tile.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] src The source tile to which the natural logarithm function is applied.
+ */
+template<typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+log(threadgroup ST &dst, threadgroup const ST &src, const int threadIdx) {
+    unary_map<base_ops::log, ST>(dst, src, threadIdx);
+}
+/**
+ * @brief Applies the absolute function to each element of the source tile and stores the result in the destination tile.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] src The source tile to which the absolute function is applied.
+ */
+template<typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+abs(threadgroup ST &dst, threadgroup const ST &src, const int threadIdx) {
+    unary_map<base_ops::abs, ST>(dst, src, threadIdx);
+}
+/**
+ * @brief Applies the rectified linear unit function to each element of the source tile and stores the result in the destination tile.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] src The source tile to which the rectified linear unit function is applied.
+ */
+template<typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+relu(threadgroup ST &dst, threadgroup const ST &src, const int threadIdx) {
+    unary_map<base_ops::relu, ST>(dst, src, threadIdx);
+}
+/**
+ * @brief Copies the elements of the source tile to the destination tile.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @tparam U The type of the source data. Must be convertible to the data type of the destination tile.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] src The source data to be copied.
+ */
+template<typename ST, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ copy(threadgroup ST &dst, thread const U &src, const int threadIdx) {
+    bin_map<base_ops::copy, ST>(dst, src, threadIdx);
+}
+
+// uniform binary maps
+/**
+ * @brief Finds the maximum of each pair of corresponding elements in the two source tiles and stores the result in the destination tile.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @tparam U The type of the second source data. Must be convertible to the data type of the destination tile.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] lhs The first source tile.
+ * @param[in] rhs The second source data.
+ */
+template<typename ST, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ max(threadgroup ST &dst, threadgroup const ST &lhs, thread const U &rhs, const int threadIdx) {
+    bin_map<base_ops::max, ST>(dst, lhs, rhs, threadIdx);
+}
+/**
+ * @brief Finds the minimum of each pair of corresponding elements in the two source tiles and stores the result in the destination tile.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @tparam U The type of the second source data. Must be convertible to the data type of the destination tile.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] lhs The first source tile.
+ * @param[in] rhs The second source data.
+ */
+template<typename ST, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ min(threadgroup ST &dst, threadgroup const ST &lhs, thread const U &rhs, const int threadIdx) {
+    bin_map<base_ops::min, ST>(dst, lhs, rhs, threadIdx);
+}
+/**
+ * @brief Adds each pair of corresponding elements in the two source tiles and stores the result in the destination tile.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @tparam U The type of the second source data. Must be convertible to the data type of the destination tile.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] lhs The first source tile.
+ * @param[in] rhs The second source data.
+ */
+template<typename ST, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ add(threadgroup ST &dst, threadgroup const ST &lhs, thread const U &rhs, const int threadIdx) {
+    bin_map<base_ops::sum, ST>(dst, lhs, rhs, threadIdx);
+}
+/**
+ * @brief Subtracts each pair of corresponding elements in the two source tiles and stores the result in the destination tile.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @tparam U The type of the second source data. Must be convertible to the data type of the destination tile.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] lhs The first source tile.
+ * @param[in] rhs The second source data.
+ */
+template<typename ST, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ sub(threadgroup ST &dst, threadgroup const ST &lhs, thread const U &rhs, const int threadIdx) {
+    bin_map<base_ops::sub, ST>(dst, lhs, rhs, threadIdx);
+}
+/**
+ * @brief Multiplies each pair of corresponding elements in the two source tiles and stores the result in the destination tile.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @tparam U The type of the second source data. Must be convertible to the data type of the destination tile.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] lhs The first source tile.
+ * @param[in] rhs The second source data.
+ */
+template<typename ST, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+ mul(threadgroup ST &dst, threadgroup const ST &lhs, thread const U &rhs, const int threadIdx) {
+    bin_map<base_ops::mul, ST>(dst, lhs, rhs, threadIdx);
+}
+/**
+ * @brief Divides each pair of corresponding elements in the two source tiles and stores the result in the destination tile.
+ *
+ * @tparam T The type of the tile. Must satisfy the `ducks::st::all` concept.
+ * @tparam U The type of the second source data. Must be convertible to the data type of the destination tile.
+ * @param[out] dst The destination tile where the results are stored.
+ * @param[in] lhs The first source tile.
+ * @param[in] rhs The second source data.
+ */
+template<typename ST, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>(), void>::type
+div(threadgroup ST &dst, threadgroup const ST &lhs, thread const U &rhs, const int threadIdx) {
+    bin_map<base_ops::div, ST>(dst, lhs, rhs, threadIdx);
+}
+
+// Row and col maps
+
+/**
+ * @brief Adds row values to each row of a tile.
+ *
+ * @tparam T Tile type.
+ * @tparam V Column vector type.
+ * @param dst[out] Destination tile where the result is stored.
+ * @param src[in] Source tile to apply the addition on.
+ * @param row_values[in] Column vector containing values to add to each row.
+ */
+template<typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ add_row(threadgroup ST &dst, threadgroup const ST &src, threadgroup const SV &row_values, const int threadIdx) {
+    row_map<base_ops::sum, ST, SV>(dst, src, row_values, threadIdx);
+}
+/**
+ * @brief Subtracts row values from each row of a tile.
+ *
+ * @tparam T Tile type.
+ * @tparam V Column vector type.
+ * @param dst[out] Destination tile where the result is stored.
+ * @param src[in] Source tile to apply the subtraction on.
+ * @param row_values[in] Column vector containing values to subtract from each row.
+ */
+template<typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ sub_row(threadgroup ST &dst, threadgroup const ST &src, threadgroup const SV &row_values, const int threadIdx) {
+    row_map<base_ops::sub, ST, SV>(dst, src, row_values, threadIdx);
+}
+/**
+ * @brief Multiplies each row of a tile by row values.
+ *
+ * @tparam T Tile type.
+ * @tparam V Column vector type.
+ * @param dst[out] Destination tile where the result is stored.
+ * @param src[in] Source tile to apply the multiplication on.
+ * @param row_values[in] Column vector containing values to multiply each row by.
+ */
+template<typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ mul_row(threadgroup ST &dst, threadgroup const ST &src, threadgroup const SV &row_values, const int threadIdx) {
+    row_map<base_ops::mul, ST, SV>(dst, src, row_values, threadIdx);
+}
+/**
+ * @brief Divides each row of a tile by row values.
+ *
+ * @tparam T Tile type.
+ * @tparam V Column vector type.
+ * @param dst[out] Destination tile where the result is stored.
+ * @param src[in] Source tile to apply the division on.
+ * @param row_values[in] Column vector containing values to divide each row by.
+ */
+template<typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ div_row(threadgroup ST &dst, threadgroup const ST &src, threadgroup const SV &row_values, const int threadIdx) {
+    row_map<base_ops::div, ST, SV>(dst, src, row_values, threadIdx);
+}
+/**
+ * @brief Broadcast a vector into into a tile's rows.
+ *
+ * @tparam T Tile type.
+ * @tparam V Column vector type.
+ * @param dst[out] Destination tile where the result is stored.
+ * @param row_values[in] Column vector containing values to broadcast into rows.
+ */
+template<typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ broadcast_row(threadgroup ST &dst, threadgroup const SV &row_values, const int threadIdx) {
+    row_map<base_ops::copy2, ST, SV>(dst, dst, row_values, threadIdx);
+}
+
+
+// col maps
+/**
+ * @brief Adds column values to each column of a tile.
+ *
+ * @tparam T Tile type.
+ * @tparam V Row vector type.
+ * @param dst[out] Destination tile where the result is stored.
+ * @param src[in] Source tile to apply the addition on.
+ * @param col_values[in] Row vector containing values to add to each column.
+ */
+template<typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ add_col(threadgroup ST &dst, threadgroup const ST &src, threadgroup const SV &col_values, const int threadIdx) {
+    col_map<base_ops::sum, ST, SV>(dst, src, col_values, threadIdx);
+}
+/**
+ * @brief Subtracts column values from each column of a tile.
+ *
+ * @tparam T Tile type.
+ * @tparam V Row vector type.
+ * @param dst[out] Destination tile where the result is stored.
+ * @param src[in] Source tile to apply the subtraction on.
+ * @param col_values[in] Row vector containing values to subtract from each column.
+ */
+template<typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ sub_col(threadgroup ST &dst, threadgroup const ST &src, threadgroup const SV &col_values, const int threadIdx) {
+    col_map<base_ops::sub, ST, SV>(dst, src, col_values, threadIdx);
+}
+/**
+ * @brief Multiplies each column of a tile by column values.
+ *
+ * @tparam T Tile type.
+ * @tparam V Row vector type.
+ * @param dst[out] Destination tile where the result is stored.
+ * @param src[in] Source tile to apply the multiplication on.
+ * @param col_values[in] Row vector containing values to multiply each column by.
+ */
+template<typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ mul_col(threadgroup ST &dst, threadgroup const ST &src, threadgroup const SV &col_values, const int threadIdx) {
+     col_map<base_ops::mul, ST, SV>(dst, src, col_values, threadIdx);
+}
+/**
+ * @brief Divides each column of a tile by column values.
+ *
+ * @tparam T Tile type.
+ * @tparam V Row vector type.
+ * @param dst[out] Destination tile where the result is stored.
+ * @param src[in] Source tile to apply the division on.
+ * @param col_values[in] Row vector containing values to divide each column by.
+ */
+template<typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ div_col(threadgroup ST &dst, threadgroup const ST &src, threadgroup const SV &col_values, const int threadIdx) {
+    col_map<base_ops::div, ST, SV>(dst, src, col_values, threadIdx);
+}
+/**
+ * @brief Broadcast a vector into into a tile's columns.
+ *
+ * @tparam T Tile type.
+ * @tparam V Row vector type.
+ * @param dst[out] Destination tile where the result is stored.
+ * @param row_values[in] Row vector containing values to broadcast into cols.
+ */
+template<typename ST, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ broadcast_col(threadgroup ST &dst, threadgroup const SV &col_values, const int threadIdx) {
+    col_map<base_ops::copy2, ST, SV>(dst, dst, col_values, threadIdx);
+}
@@ -0,0 +1,284 @@
+/**
+ * @file
+ * @brief Group reductions on shared tiles.
+ */
+
+/**
+ * Performs row-wise reduction on a matrix using a specified operation.
+ *
+ * @tparam op The operation to be applied for reduction.
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type with row layout.
+ * @param row_accum The accumulator where the result of the reduction is stored.
+ * @param src The source matrix on which to perform the reduction.
+ * @param src_accum The initial value of the accumulator, used when reset is false.
+ * @param reset A boolean flag indicating whether to reset the accumulator (ignore src_accum) or not.
+ */
+template<typename op, typename SV, typename ST, bool reset>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ row_reduce(threadgroup SV &row_accum, threadgroup const ST &src, threadgroup const SV &src_accum, const int threadIdx) {
+    using dtype = typename SV::dtype;
+    for (int row = laneid(threadIdx); row < src.rows; row += GROUP_THREADS) {
+        dtype accum = src[{row, 0}];
+        #pragma clang loop unroll(full)
+        for (int col = 1; col < src.cols; col++) {
+            accum = op::template op<dtype>(accum, src[{row, col}]);
+        }
+        if (reset) {
+            row_accum[row] = accum;
+        } else {
+            row_accum[row] = op::template op<dtype>(src_accum[row], accum);
+        }
+    }
+}
+
+/**
+ * Performs column-wise reduction on a matrix using a specified operation.
+ *
+ * @tparam op The operation to be applied for reduction.
+ * @tparam V The shared vector type for the column accumulator.
+ * @tparam T The shared matrix type with column layout.
+ * @param col_accum The accumulator where the result of the reduction is stored.
+ * @param src The source matrix on which to perform the reduction.
+ * @param src_accum The initial value of the accumulator, used when reset is false.
+ * @param reset A boolean flag indicating whether to reset the accumulator (ignore src_accum) or not.
+ */
+template<typename op, typename SV, typename ST, bool reset>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ col_reduce(threadgroup SV &col_accum, threadgroup const ST &src, threadgroup const SV &src_accum, const int threadIdx) {
+    using dtype = typename SV::dtype;
+    for (int col = laneid(threadIdx); col < src.cols; col += GROUP_THREADS) {
+        dtype accum = src[{0, col}];
+        #pragma clang loop unroll(full)
+        for (int row = 1; row < src.rows; row++) {
+            accum = op::template op<dtype>(accum, src[{row, col}]);
+        }
+        if (reset) {
+            col_accum[col] = accum;
+        } else {
+            col_accum[col] = op::template op<dtype>(src_accum[col], accum);
+        }
+    }
+}
+
+/* ----------  WRAPPERS FOR PRETTINESS  ---------- */
+
+/**
+ * @brief Store the maximum of each row of the src shared matrix in the row_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] row_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ row_max(threadgroup SV &row_accum, threadgroup const ST &src, const int threadIdx)  {
+    row_reduce<base_ops::max, SV, ST, true>(row_accum, src, row_accum, threadIdx);
+}
+/**
+ * @brief Store the minimum of each row of the src shared matrix in the row_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] row_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ row_min(threadgroup SV &row_accum, threadgroup const ST &src, const int threadIdx)  {
+    row_reduce<base_ops::min, SV, ST, true>(row_accum, src, row_accum, threadIdx);
+}
+/**
+ * @brief Store the sum of each row of the src shared matrix in the row_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] row_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ row_sum(threadgroup SV &row_accum, threadgroup const ST &src, const int threadIdx)  {
+    row_reduce<base_ops::sum, SV, ST, true>(row_accum, src, row_accum, threadIdx);
+}
+/**
+ * @brief Store the product of each row of the src shared matrix in the row_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] row_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ row_prod(threadgroup SV &row_accum, threadgroup const ST &src, const int threadIdx) {
+    row_reduce<base_ops::mul, SV, ST, true>(row_accum, src, row_accum, threadIdx);
+}
+
+/**
+ * @brief Store the maximum of each row of the src shared matrix, as well as the src_accum shared vector, in the row_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] row_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ * @param[in] src_accum The initial value of the accumulator, used when accumulating onto an existing value.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ row_max(threadgroup SV &row_accum, threadgroup const ST &src, threadgroup const SV &src_accum, const int threadIdx)  {
+    row_reduce<base_ops::max, SV, ST, false>(row_accum, src, src_accum, threadIdx);
+}
+/**
+ * @brief Store the minimum of each row of the src shared matrix, as well as the src_accum shared vector, in the row_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] row_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ * @param[in] src_accum The initial value of the accumulator, used when accumulating onto an existing value.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ row_min(threadgroup SV &row_accum, threadgroup const ST &src, threadgroup const SV &src_accum, const int threadIdx)  {
+    row_reduce<base_ops::min, SV, ST, false>(row_accum, src, src_accum, threadIdx);
+}
+/**
+ * @brief Store the sum of each row of the src shared matrix, as well as the src_accum shared vector, in the row_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] row_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ * @param[in] src_accum The initial value of the accumulator, used when accumulating onto an existing value.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ row_sum(threadgroup SV &row_accum, threadgroup const ST &src, threadgroup const SV &src_accum, const int threadIdx)  {
+    row_reduce<base_ops::sum, SV, ST, false>(row_accum, src, src_accum, threadIdx);
+}
+/**
+ * @brief Store the product of each row of the src shared matrix, as well as the src_accum shared vector, in the row_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] row_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ * @param[in] src_accum The initial value of the accumulator, used when accumulating onto an existing value.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ row_prod(threadgroup SV &row_accum, threadgroup const ST &src, threadgroup const SV &src_accum, const int threadIdx) {
+    row_reduce<base_ops::mul, SV, ST, false>(row_accum, src, src_accum, threadIdx);
+}
+
+/**
+ * @brief Store the maximum of each column of the src shared matrix in the col_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] col_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ col_max(threadgroup SV &col_accum, threadgroup const ST &src, const int threadIdx)  {
+    col_reduce<base_ops::max, SV, ST, true>(col_accum, src, col_accum, threadIdx);
+}
+/**
+ * @brief Store the minimum of each column of the src shared matrix in the col_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] col_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ col_min(threadgroup SV &col_accum, threadgroup const ST &src, const int threadIdx)  {
+    col_reduce<base_ops::min, threadgroup SV, threadgroup ST, true>(col_accum, src, col_accum, threadIdx);
+}
+/**
+ * @brief Store the sum of each column of the src shared matrix in the col_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] col_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ col_sum(threadgroup SV &col_accum, threadgroup const ST &src, const int threadIdx)  {
+    col_reduce<base_ops::sum, SV, ST, true>(col_accum, src, col_accum, threadIdx);
+}
+/**
+ * @brief Store the product of each column of the src shared matrix in the col_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] col_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ col_prod(threadgroup SV &col_accum, threadgroup const ST &src, const int threadIdx) {
+    col_reduce<base_ops::mul, SV, ST, true>(col_accum, src, col_accum, threadIdx);
+}
+
+/**
+ * @brief Store the maximum of each column of the src shared matrix, as well as the src_accum shared vector, in the col_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] col_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ * @param[in] src_accum The initial value of the accumulator, used when accumulating onto an existing value.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ col_max(threadgroup SV &col_accum, threadgroup const ST &src, threadgroup const SV &src_accum, const int threadIdx)  {
+    col_reduce<base_ops::max, SV, ST, false>(col_accum, src, src_accum, threadIdx);
+}
+/**
+ * @brief Store the minimum of each column of the src shared matrix, as well as the src_accum shared vector, in the col_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The matrix type.
+ * @param[out] col_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ * @param[in] src_accum The initial value of the accumulator, used when accumulating onto an existing value.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ col_min(threadgroup SV &col_accum, threadgroup const ST &src, threadgroup const SV &src_accum, const int threadIdx)  {
+    col_reduce<base_ops::min, SV, ST, false>(col_accum, src, src_accum, threadIdx);
+}
+/**
+ * @brief Store the sum of each column of the src shared tile, as well as the src_accum row vector, in the col_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] col_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ * @param[in] src_accum The initial value of the accumulator, used when accumulating onto an existing value.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ col_sum(threadgroup SV &col_accum, threadgroup const ST &src, threadgroup const SV &src_accum, const int threadIdx)  {
+    col_reduce<base_ops::sum, SV, ST, false>(col_accum, src, src_accum, threadIdx);
+}
+/**
+ * @brief Store the product of each column of the src shared tile, as well as the src_accum row vector, in the col_accum shared vector.
+ *
+ * @tparam V The shared vector type for the row accumulator.
+ * @tparam T The shared matrix type.
+ * @param[out] col_accum The accumulator where the result of the reduction is stored.
+ * @param[in] src The source matrix on which to perform the reduction.
+ * @param[in] src_accum The initial value of the accumulator, used when accumulating onto an existing value.
+ */
+template<typename SV, typename ST>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_tile<ST>() && ducks::is_shared_vector<SV>(), void>::type
+ col_prod(threadgroup SV &col_accum, threadgroup const ST &src, threadgroup const SV &src_accum, const int threadIdx) {
+    col_reduce<base_ops::mul, SV, ST, false>(col_accum, src, src_accum, threadIdx);
+}
@@ -0,0 +1,3 @@
+#include "conversions.metal"
+#include "maps.metal"
+#include "reductions.metal"
@@ -0,0 +1,29 @@
+/**
+ * @file
+ * @brief Group conversions on shared vectors.
+ */
+
+/**
+ * @brief Copies data from one shared vector to another, converting data types if necessary.
+ *
+ * This function copies data from the source shared vector `src` to the destination shared vector `dst`.
+ * If the data types of `src` and `dst` are the same, it performs a direct memory copy. Otherwise, it
+ * converts each element from the source data type to the destination data type using the appropriate
+ * converter before copying.
+ *
+ * @tparam SV1 The type of the destination shared vector, must satisfy the ducks::sv::all concept.
+ * @tparam SV2 The type of the source shared vector, must satisfy the ducks::sv::all concept.
+ * @param[out] dst The destination shared vector.
+ * @param[in] src The source shared vector.
+ * @note The lengths of `src` and `dst` must be equal. This is enforced at compile time.
+ */
+template<typename SV1, typename SV2>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV1>() && ducks::is_shared_vector<SV2>(), void>::type
+copy(threadgroup SV1 &dst, threadgroup const SV2 &src, const int threadIdx) {
+    static_assert(SV1::length == SV2::length, "Source and destination vectors must have the same length.");
+    #pragma clang loop unroll(full)
+    for(int i = laneid(threadIdx); i < dst.length; i+=GROUP_THREADS) {
+        dst[i] = base_types::convertor<typename SV1::dtype, typename SV2::dtype>::convert(src[i]);
+    }
+}
+
@@ -0,0 +1,267 @@
+/**
+ * @file
+ * @brief Group maps on shared vectors.
+ */
+
+/**
+ * @brief Applies a unary operation to each element of a shared memory vector.
+ *
+ * @tparam op Unary operation type.
+ * @tparam T Shared memory vector type.
+ * @param dst[out] Destination vector in which to store the result.
+ * @param src[in] Source vector to apply the unary operation.
+ */
+template<typename op, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+unary_op(threadgroup SV &dst, threadgroup const SV &src, const int threadIdx) {
+    #pragma clang loop unroll(full)
+    for(auto cur = laneid(threadIdx); cur < SV::length; cur+=GROUP_THREADS) {
+        dst[cur] = op::template op<typename SV::dtype>(src[cur]);
+    }
+}
+/**
+ * @brief Perform a binary operation on two shared vectors.
+ *
+ * @tparam op The binary operation to perform.
+ * @tparam T The type of the vectors.
+ * @param dst[out] The destination vector where the result is stored.
+ * @param lhs[in] The left-hand side vector for the operation.
+ * @param rhs[in] The right-hand side vector for the operation.
+ */
+template<typename op, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+bin_op(threadgroup SV &dst, threadgroup const SV &lhs, threadgroup const SV &rhs, const int threadIdx) {
+    #pragma clang loop unroll(full)
+    for(auto cur = laneid(threadIdx); cur < SV::length; cur+=GROUP_THREADS) {
+        dst[cur] = op::template op<typename SV::dtype>(lhs[cur], rhs[cur]);
+    }
+}
+/**
+ * @brief Perform a binary operation on a shared vector and a scalar.
+ *
+ * @tparam op The binary operation to perform.
+ * @tparam T The type of the vector.
+ * @param dst[out] The destination vector where the result is stored.
+ * @param src[in] The source vector for the operation.
+ * @param param[in] The scalar parameter for the operation.
+ */
+template<typename op, typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+bin_op(threadgroup SV &dst, threadgroup const SV &src, thread const typename SV::dtype &param, const int threadIdx) {
+    #pragma clang loop unroll(full)
+    for(auto cur = laneid(threadIdx); cur < SV::length; cur+=GROUP_THREADS) {
+        dst[cur] = op::template op<typename SV::dtype>(src[cur], param);
+    }
+}
+
+/* ----------  WRAPPERS FOR PRETTINESS  ---------- */
+
+// ---- const ops ----
+
+/**
+ * @brief Sets all elements of a shared memory vector to zero.
+ *
+ * @tparam T Shared memory vector type.
+ * @param dst[out] Destination vector to be set to zero.
+ */
+template<typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+zero(threadgroup SV &dst, const int threadIdx) {
+    unary_op<base_ops::zero, SV>(dst, dst, threadIdx);
+}
+/**
+ * @brief Sets all elements of a shared memory vector to one.
+ *
+ * @tparam T Shared memory vector type.
+ * @param dst[out] Destination vector to be set to one.
+ */
+template<typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+one(threadgroup SV &dst, const int threadIdx) {
+    unary_op<base_ops::one, SV>(dst, dst, threadIdx);
+}
+/**
+ * @brief Sets all elements of a shared memory vector to positive infinity.
+ *
+ * @tparam T Shared memory vector type.
+ * @param dst[out] Destination vector to be set to positive infinity.
+ */
+template<typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+pos_infty(threadgroup SV &dst, const int threadIdx) {
+    unary_op<base_ops::pos_infty, SV>(dst, dst, threadIdx);
+}
+/**
+ * @brief Sets all elements of a shared memory vector to negative infinity.
+ *
+ * @tparam T Shared memory vector type.
+ * @param dst[out] Destination vector to be set to negative infinity.
+ */
+template<typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+neg_infty(threadgroup SV &dst, const int threadIdx) {
+    unary_op<base_ops::neg_infty, SV>(dst, dst, threadIdx);
+}
+
+// ---- unary ops ----
+
+/**
+ * @brief Copies the elements from one shared vector to another.
+ *
+ * @tparam T Shared vector type.
+ * @tparam U Type of the source vector.
+ * @param dst[out] Destination vector where the elements will be copied to.
+ * @param src[in] Source vector to copy the elements from.
+ */
+template<typename SV, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+copy(threadgroup SV &dst, thread const U &src, const int threadIdx) {
+    bin_op<base_ops::copy2, SV>(dst, dst, src, threadIdx); // the second arg is ignored here.
+}
+/**
+ * @brief Applies the exponential function element-wise to a shared vector.
+ *
+ * @tparam T Shared vector type.
+ * @param dst[out] Destination vector where the exponential values will be stored.
+ * @param src[in] Source vector to apply the exponential function to.
+ */
+template<typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+exp(threadgroup SV &dst, threadgroup const SV &src, const int threadIdx) {
+    unary_op<base_ops::exp, SV>(dst, src, threadIdx);
+}
+/**
+ * @brief Applies the exponential function element-wise to a shared vector, in base 2.
+ *
+ * @tparam T Shared vector type.
+ * @param dst[out] Destination vector where the exponential values will be stored.
+ * @param src[in] Source vector to apply the exponential function to.
+ */
+template<typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+exp2(threadgroup SV &dst, threadgroup const SV &src, const int threadIdx) {
+    unary_op<base_ops::exp2, SV>(dst, src, threadIdx);
+}
+/**
+ * @brief Applies the natural logarithm function element-wise to a shared vector.
+ *
+ * @tparam T Shared vector type.
+ * @param dst[out] Destination vector where the exponential values will be stored.
+ * @param src[in] Source vector to apply the logarithm function to.
+ */
+template<typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+log(threadgroup SV &dst, threadgroup const SV &src, const int threadIdx) {
+    unary_op<base_ops::log, SV>(dst, src, threadIdx);
+}
+/**
+ * @brief Applies the absolute value function element-wise to a shared vector.
+ *
+ * @tparam T Shared vector type.
+ * @param dst[out] Destination vector where the absolute values will be stored.
+ * @param src[in] Source vector to apply the absolute value function to.
+ */
+template<typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+abs(threadgroup SV &dst, threadgroup const SV &src, const int threadIdx) {
+    unary_op<base_ops::abs, SV>(dst, src, threadIdx);
+}
+/**
+ * @brief Applies the rectified linear unit (ReLU) function element-wise to a shared vector.
+ *
+ * @tparam T Shared vector type.
+ * @param dst[out] Destination vector where the ReLU values will be stored.
+ * @param src[in] Source vector to apply the ReLU function to.
+ */
+template<typename SV>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+relu(threadgroup SV &dst, threadgroup const SV &src, const int threadIdx) {
+    unary_op<base_ops::relu, SV>(dst, src, threadIdx);
+}
+
+// ---- binary ops ----
+
+/**
+ * @brief Computes the element-wise maximum of two shared vectors.
+ *
+ * @tparam T Shared vector type.
+ * @tparam U Type of the second vector.
+ * @param dst[out] Destination vector where the maximum values will be stored.
+ * @param lhs[in] First vector for the maximum operation.
+ * @param rhs[in] Second vector for the maximum operation.
+ */
+template<typename SV, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+max(threadgroup SV &dst, threadgroup const SV &lhs, thread const U &rhs, const int threadIdx) {
+    bin_op<base_ops::max, SV>(dst, lhs, rhs, threadIdx);
+}
+/**
+ * @brief Computes the element-wise minimum of two shared vectors.
+ *
+ * @tparam T Shared vector type.
+ * @tparam U Type of the second vector.
+ * @param dst[out] Destination vector where the minimum values will be stored.
+ * @param lhs[in] First vector for the minimum operation.
+ * @param rhs[in] Second vector for the minimum operation.
+ */
+template<typename SV, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+min(threadgroup SV &dst, threadgroup const SV &lhs, thread const U &rhs, const int threadIdx) {
+    bin_op<base_ops::min, SV>(dst, lhs, rhs, threadIdx);
+}
+/**
+ * @brief Computes the element-wise sum of two shared vectors.
+ *
+ * @tparam T Shared vector type.
+ * @tparam U Type of the second vector.
+ * @param dst[out] Destination vector where the sum values will be stored.
+ * @param lhs[in] First vector for the sum operation.
+ * @param rhs[in] Second vector for the sum operation.
+ */
+template<typename SV, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+add(threadgroup SV &dst, threadgroup const SV &lhs, thread const U &rhs, const int threadIdx) {
+    bin_op<base_ops::sum, SV>(dst, lhs, rhs, threadIdx);
+}
+/**
+ * @brief Computes the element-wise difference of two shared vectors.
+ *
+ * @tparam T Shared vector type.
+ * @tparam U Type of the second vector.
+ * @param dst[out] Destination vector where the difference values will be stored.
+ * @param lhs[in] First vector for the difference operation.
+ * @param rhs[in] Second vector for the difference operation.
+ */
+template<typename SV, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+sub(threadgroup SV &dst, threadgroup const SV &lhs, thread const U &rhs, const int threadIdx) {
+    bin_op<base_ops::sub, SV>(dst, lhs, rhs, threadIdx);
+}
+/**
+ * @brief Computes the element-wise product of two shared vectors.
+ *
+ * @tparam T Shared vector type.
+ * @tparam U Type of the second vector.
+ * @param dst[out] Destination vector where the product values will be stored.
+ * @param lhs[in] First vector for the product operation.
+ * @param rhs[in] Second vector for the product operation.
+ */
+template<typename SV, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+mul(threadgroup SV &dst, threadgroup const SV &lhs, thread const U &rhs, const int threadIdx) {
+    bin_op<base_ops::mul, SV>(dst, lhs, rhs, threadIdx);
+}
+/**
+ * @brief Computes the element-wise division of two shared vectors.
+ *
+ * @tparam T Shared vector type.
+ * @tparam U Type of the second vector.
+ * @param dst[out] Destination vector where the division values will be stored.
+ * @param lhs[in] First vector for the division operation.
+ * @param rhs[in] Second vector for the division operation.
+ */
+template<typename SV, typename U>
+static METAL_FUNC typename metal::enable_if<ducks::is_shared_vector<SV>(), void>::type
+div(threadgroup SV &dst, threadgroup const SV &lhs, thread const U &rhs, const int threadIdx) {
+    bin_op<base_ops::div, SV>(dst, lhs, rhs, threadIdx);
+}
@@ -0,0 +1,3 @@
+#include "conversions.metal"
+#include "maps.metal"
+
--- a/Show More
+++ b/Show More