tinygrad/tinygrad
2
0
Fork 3
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-08-29 12:56:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'master' into shrink_in_render

Browse Source
This commit is contained in:
George Hotz
2026-06-01 19:10:41 -07:00
committed by GitHub
parent 1ffe08bab9 c6cad1ad67
commit 2bcd46d946
1 changed files with 41 additions and 41 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/test.yml
+41 -41
View File
@@ -143,61 +143,61 @@ jobs:
with:
key: be-minimal
deps: testing_unit
- name: Test dtype with Python emulator
run: DEBUG=1 DEV=PYTHON python3 -m pytest -n=auto test/backend/test_dtype.py test/backend/test_dtype_alu.py
- name: Test ops with Python emulator
run: DEBUG=2 SKIP_SLOW_TEST=1 DEV=PYTHON python3 -m pytest -n=auto test/backend/test_ops.py --durations=20
- name: Test uops with Python emulator
run: DEV=PYTHON python3 -m pytest test/backend/test_uops.py --durations=20
- name: Test symbolic with Python emulator
run: DEV=PYTHON python3 test/backend/test_symbolic_ops.py
- name: test_renderer_failures with Python emulator
run: DEV=PYTHON python3 -m pytest -rA test/backend/test_renderer_failures.py::TestRendererFailures
- name: Run backend tests
run: SKIP_SLOW_TEST=1 DEV=PYTHON python3 -m pytest -n=auto test/backend/test_dtype.py test/backend/test_dtype_alu.py test/backend/test_ops.py test/backend/test_uops.py test/backend/test_symbolic_ops.py test/backend/test_renderer_failures.py::TestRendererFailures --durations=20
- name: Test IMAGE support
run: |
IMAGE=1 DEV=PYTHON python3 test/backend/test_ops.py TestOps.test_gemm
IMAGE=1 DEV=PYTHON python3 test/backend/test_ops.py TestOps.test_simple_conv2d
run: IMAGE=1 DEV=PYTHON python3 test/backend/test_ops.py TestOps.test_gemm TestOps.test_simple_conv2d
- name: Test emulated METAL tensor cores
env:
DEV: 'PYTHON::METAL'
run: |
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::METAL python3 test/backend/test_ops.py TestOps.test_big_gemm
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::METAL python3 test/opt/test_tensor_cores.py
- name: Test emulated AMX tensor cores
run: DEBUG=2 AMX=1 FORWARD_ONLY=1 DEV=PYTHON::AMX python3 test/backend/test_ops.py TestOps.test_gemm
DEBUG=2 python3 test/backend/test_ops.py TestOps.test_big_gemm
python3 -m pytest -nauto test/opt/test_tensor_cores.py
- name: Test emulated AMD tensor cores
env:
DEV: 'PYTHON::gfx1100'
run: |
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx1100 N=16 HALF=1 ACC_HALF=0 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx1100 N=64 HALF=1 ACC_HALF=0 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx1100 N=16 HALF=1 ACC_HALF=1 ATOL=1e-3 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx1100 N=64 HALF=1 ACC_HALF=1 ATOL=1e-3 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx1100 python3 test/opt/test_tensor_cores.py
DEBUG=2 N=16 HALF=1 ACC_HALF=0 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 N=64 HALF=1 ACC_HALF=0 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 N=16 HALF=1 ACC_HALF=1 ATOL=1e-3 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 N=64 HALF=1 ACC_HALF=1 ATOL=1e-3 python3 ./extra/gemm/simple_matmul.py
python3 -m pytest -nauto test/opt/test_tensor_cores.py
- name: Test emulated AMD MFMA tensor cores
env:
DEV: 'PYTHON::gfx950'
run: |
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx950 N=64 HALF=1 ACC_HALF=0 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx950 python3 test/opt/test_tensor_cores.py
DEBUG=2 N=64 HALF=1 ACC_HALF=0 python3 ./extra/gemm/simple_matmul.py
python3 -m pytest -nauto test/opt/test_tensor_cores.py
- name: Test emulated AMD RDNA4 tensor cores
env:
DEV: 'PYTHON::gfx1201'
run: |
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx1201 N=16 HALF=1 ACC_HALF=0 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx1201 N=64 HALF=1 ACC_HALF=0 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx1201 N=16 HALF=1 ACC_HALF=1 ATOL=1e-3 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx1201 N=64 HALF=1 ACC_HALF=1 ATOL=1e-3 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::gfx1201 python3 test/opt/test_tensor_cores.py
DEBUG=2 N=16 HALF=1 ACC_HALF=0 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 N=64 HALF=1 ACC_HALF=0 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 N=16 HALF=1 ACC_HALF=1 ATOL=1e-3 python3 ./extra/gemm/simple_matmul.py
DEBUG=2 N=64 HALF=1 ACC_HALF=1 ATOL=1e-3 python3 ./extra/gemm/simple_matmul.py
python3 -m pytest -nauto test/opt/test_tensor_cores.py
- name: Test emulated CUDA tensor cores
run: |
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::sm_80 python3 test/backend/test_ops.py TestOps.test_gemm_fp16
DEBUG=2 ALLOW_TF32=1 FORWARD_ONLY=1 DEV=PYTHON::sm_80 python3 test/backend/test_ops.py TestOps.test_gemm
DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::sm_75 python3 test/backend/test_ops.py TestOps.test_gemm_fp16
DEBUG=2 ALLOW_TF32=1 FORWARD_ONLY=1 DEV=PYTHON::sm_89 python3 test/opt/test_tensor_cores.py
DEBUG=2 DEV=PYTHON::sm_80 python3 test/backend/test_ops.py TestOps.test_gemm_fp16
DEBUG=2 ALLOW_TF32=1 DEV=PYTHON::sm_80 python3 test/backend/test_ops.py TestOps.test_gemm
DEBUG=2 DEV=PYTHON::sm_75 python3 test/backend/test_ops.py TestOps.test_gemm_fp16
ALLOW_TF32=1 DEV=PYTHON::sm_89 python3 -m pytest -nauto test/opt/test_tensor_cores.py
- name: Test emulated INTEL OpenCL tensor cores
run: DEBUG=2 FORWARD_ONLY=1 DEV=PYTHON::INTEL HALF=1 N=64 python3 ./extra/gemm/simple_matmul.py
run: DEBUG=2 DEV=PYTHON::INTEL HALF=1 N=64 python3 ./extra/gemm/simple_matmul.py
- name: Test emulated AMX tensor cores
run: DEBUG=2 AMX=1 FORWARD_ONLY=1 DEV=PYTHON::AMX python3 test/opt/test_tensor_cores.py
env:
DEV: 'PYTHON::AMX'
run: |
DEBUG=2 python3 test/backend/test_ops.py TestOps.test_gemm
python3 -m pytest -nauto test/opt/test_tensor_cores.py
- name: Test device flop counts
run: |
DEBUG=2 DEV=PYTHON::METAL python3 ./test/null/test_uops_stats.py TestUOpsStatsMatmulHalf
DEBUG=2 DEV=PYTHON::gfx1100 python3 ./test/null/test_uops_stats.py TestUOpsStatsMatmulHalf
DEBUG=2 DEV=PYTHON::sm_80 python3 ./test/null/test_uops_stats.py TestUOpsStatsMatmulHalf
DEBUG=2 DEV=PYTHON::INTEL python3 ./test/null/test_uops_stats.py TestUOpsStatsMatmulHalf
DEBUG=2 AMX=1 DEV=PYTHON::AMX python3 ./test/null/test_uops_stats.py TestUOpsStats.test_simple_matmul
DEBUG=2 DEV=PYTHON::AMX python3 ./test/null/test_uops_stats.py TestUOpsStats.test_simple_matmul
linter:
name: Linters
@@ -517,7 +517,7 @@ jobs:
testmetalmodels:
name: Models (metal)
runs-on: macos-14
runs-on: &macos macos-26
timeout-minutes: 20
steps:
- name: Checkout Code
@@ -749,7 +749,7 @@ jobs:
unittestmacos:
name: MacOS (unit)
runs-on: macos-14
runs-on: *macos
timeout-minutes: 20
steps:
- name: Checkout Code
@@ -805,7 +805,7 @@ jobs:
- 'WEBGPU'
name: MacOS (DEV=${{ matrix.dev }})
runs-on: macos-15
runs-on: *macos
timeout-minutes: 20
steps:
- name: Checkout Code
@@ -843,7 +843,7 @@ jobs:
- 'WEBGPU'
name: Windows (DEV=${{ matrix.dev }})
runs-on: windows-latest
runs-on: windows-2025
timeout-minutes: 15
steps:
- name: Checkout Code
@@ -906,7 +906,7 @@ jobs:
python -m pytest -n=auto test/backend/test_ops.py --durations=20
qcomclcompiletests:
name: Compile-only (QCOM CL)
runs-on: ${{ github.repository == 'tinygrad/tinygrad' && github.event_name == 'pull_request' && github.event.pull_request.author_association == 'MEMBER' && 'namespace-profile-tinygrad-arm64' || 'ubuntu-24.04-arm' }}
runs-on: ubuntu-24.04-arm
timeout-minutes: 15
steps:
- name: Checkout Code