George Hotz and GitHub
0252cb8fa7
remove anchors from CI flow ( #17521 )
...
Autogen / In-tree Autogen (macos) (push) Waiting to run
Benchmarks / Mac pytest (push) Waiting to run
Benchmarks / LLM (DEV=AMD) (push) Waiting to run
Benchmarks / LLM (DEV=METAL) (push) Waiting to run
Benchmarks / LLM (DEV=NV) (push) Waiting to run
Benchmarks / HLB-CIFAR10 (DEV=AMD) (push) Waiting to run
Benchmarks / HLB-CIFAR10 (DEV=METAL) (push) Waiting to run
Benchmarks / HLB-CIFAR10 (DEV=NV) (push) Waiting to run
Benchmarks / MLPerf (AMD) (push) Waiting to run
Benchmarks / MLPerf (NV) (push) Waiting to run
Benchmarks / Stable Diffusion (DEV=AMD) (push) Waiting to run
Benchmarks / Stable Diffusion (DEV=METAL) (push) Waiting to run
Benchmarks / Stable Diffusion (DEV=NV) (push) Waiting to run
Benchmarks / Tests (DEV=AMD) (push) Waiting to run
Benchmarks / Tests (DEV=METAL) (push) Waiting to run
Benchmarks / Tests (DEV=NV) (push) Waiting to run
Benchmarks / UsbGPU Benchmark (push) Waiting to run
Benchmarks / comma Benchmark (0.11.2) (push) Waiting to run
Benchmarks / comma Benchmark (0.11.0) (push) Waiting to run
Benchmarks / DSP Benchmark (push) Waiting to run
Benchmarks / UsbGPU Benchmark (comma) (push) Waiting to run
Benchmarks / PCI Driver Benchmark (DEV=AMD) (push) Waiting to run
Benchmarks / PCI Driver Benchmark (DEV=NV) (push) Waiting to run
Benchmarks / LLVM Speed (push) Waiting to run
Platform Tests / MacOS (unit) (push) Waiting to run
Platform Tests / MacOS (unit, mock) (push) Waiting to run
Platform Tests / MacOS (DEV=METAL) (1) (push) Waiting to run
Platform Tests / MacOS (DEV=METAL) (2) (push) Waiting to run
Platform Tests / MacOS (DEV=CPU:CLANG) (push) Waiting to run
Platform Tests / MacOS (DEV=CPU:LLVM) (push) Waiting to run
Platform Tests / MacOS (DEV=CPU:LVP) (push) Waiting to run
Platform Tests / MacOS (DEV=WEBGPU) (push) Waiting to run
Platform Tests / Windows (DEV=CPU:CLANG) (push) Waiting to run
Platform Tests / Windows (DEV=CPU:LLVM) (push) Waiting to run
Platform Tests / Windows (DEV=CPU:X86) (push) Waiting to run
Platform Tests / Windows (DEV=WEBGPU) (push) Waiting to run
Autogen / In-tree Autogen (comgr 2) (push) Successful in 41s
Deploy Docs / deploy (push) Successful in 46s
Autogen / In-tree Autogen (push) Successful in 47s
Unit Tests / Linters (push) Successful in 2m23s
Unit Tests / Docs (push) Successful in 3m32s
Unit Tests / Python Backend (push) Successful in 3m50s
Unit Tests / Unit Tests (push) Successful in 3m55s
Unit Tests / SPEC=2 (2) (push) Successful in 4m10s
Unit Tests / Null Tests (push) Successful in 4m25s
Unit Tests / SPEC=2 (1) (push) Successful in 4m28s
Unit Tests / Torch Backend Tests (push) Successful in 4m47s
Unit Tests / Fuzzing (push) Successful in 2m32s
Unit Tests / Linux (DSP) (push) Successful in 2m1s
Unit Tests / Models (push) Successful in 2m30s
Unit Tests / CL IMAGE Tests (push) Successful in 3m29s
Unit Tests / Test LLM (push) Successful in 2m40s
Unit Tests / openpilot Compile Tests (push) Successful in 4m25s
Unit Tests / Linux (DEV=CL) (push) Successful in 4m41s
Unit Tests / Linux (DEV=CPU:LVP) (push) Successful in 3m14s
Unit Tests / ONNX (CPU) Tests (push) Successful in 6m24s
Unit Tests / Optimization Tests (push) Successful in 6m13s
Unit Tests / Linux (DEV=CPU:LLVM) (push) Successful in 3m32s
Unit Tests / Linux (DEV=CPU:X86) (push) Successful in 3m33s
Unit Tests / Linux (DEV=CPU:CLANG) (push) Successful in 4m30s
Unit Tests / AMD ASM IDE (push) Successful in 2m44s
Unit Tests / Linux (DEV=WEBGPU) (push) Successful in 4m16s
Unit Tests / hcq2 (push) Successful in 3m13s
Unit Tests / Linux (amdllvm gfx1100) (push) Successful in 3m55s
Unit Tests / Linux (amd gfx1100) (push) Successful in 4m24s
Unit Tests / Linux (amd gfx1201) (push) Successful in 4m23s
Unit Tests / Linux (am) (push) Successful in 4m45s
Unit Tests / Linux (amdllvm gfx1201) (push) Successful in 4m6s
Unit Tests / Compile-only (DEV=NULL:QCOMCL:a630) (push) Failing after 24s
Unit Tests / Linux (amdllvm gfx950) (push) Successful in 3m51s
Unit Tests / Compile-only (DEV=NULL:NAK:sm_120) (push) Successful in 1m39s
Unit Tests / Compile-only (DEV=NULL:IR3:a630) (push) Successful in 2m20s
Unit Tests / Linux (amd gfx950) (push) Successful in 5m42s
Unit Tests / Linux (ptx) (push) Successful in 4m1s
Unit Tests / Linux (nv) (push) Successful in 5m13s
* ci: remove yaml anchors from test.yml for gitea actions compatibility
Gitea Actions does not support YAML anchors/aliases, which causes the
workflow to fail parsing. Replace the &linux/*linux anchor with a plain
runs-on: ubuntu-24.04.
* ci: keep runner selection, inline expression instead of anchors
Instead of replacing the anchored runs-on with a plain ubuntu-24.04
(which drops the namespace-profile-tinygrad routing for collaborator
PRs), inline the full ${{ }} expression at every job. No YAML anchors,
works with runners that can't parse them (gitea runner), and identical
behavior on GitHub Actions.
2026-08-13 10:17:22 -07:00
George Hotz and GitHub
e103fb2a10
more lil llm improvements ( #17514 )
...
Autogen / In-tree Autogen (macos) (push) Waiting to run
Benchmarks / Mac pytest (push) Waiting to run
Benchmarks / LLM (DEV=AMD) (push) Waiting to run
Benchmarks / LLM (DEV=METAL) (push) Waiting to run
Benchmarks / LLM (DEV=NV) (push) Waiting to run
Benchmarks / HLB-CIFAR10 (DEV=AMD) (push) Waiting to run
Benchmarks / HLB-CIFAR10 (DEV=METAL) (push) Waiting to run
Benchmarks / HLB-CIFAR10 (DEV=NV) (push) Waiting to run
Benchmarks / MLPerf (AMD) (push) Waiting to run
Benchmarks / MLPerf (NV) (push) Waiting to run
Benchmarks / Stable Diffusion (DEV=AMD) (push) Waiting to run
Benchmarks / Stable Diffusion (DEV=METAL) (push) Waiting to run
Benchmarks / Stable Diffusion (DEV=NV) (push) Waiting to run
Benchmarks / Tests (DEV=AMD) (push) Waiting to run
Benchmarks / Tests (DEV=METAL) (push) Waiting to run
Benchmarks / Tests (DEV=NV) (push) Waiting to run
Benchmarks / UsbGPU Benchmark (push) Waiting to run
Benchmarks / comma Benchmark (0.11.2) (push) Waiting to run
Benchmarks / comma Benchmark (0.11.0) (push) Waiting to run
Benchmarks / DSP Benchmark (push) Waiting to run
Benchmarks / UsbGPU Benchmark (comma) (push) Waiting to run
Benchmarks / PCI Driver Benchmark (DEV=AMD) (push) Waiting to run
Benchmarks / PCI Driver Benchmark (DEV=NV) (push) Waiting to run
Benchmarks / LLVM Speed (push) Waiting to run
Platform Tests / MacOS (unit) (push) Waiting to run
Platform Tests / MacOS (unit, mock) (push) Waiting to run
Platform Tests / MacOS (DEV=METAL) (1) (push) Waiting to run
Platform Tests / MacOS (DEV=METAL) (2) (push) Waiting to run
Platform Tests / MacOS (DEV=CPU:CLANG) (push) Waiting to run
Platform Tests / MacOS (DEV=CPU:LLVM) (push) Waiting to run
Platform Tests / MacOS (DEV=CPU:LVP) (push) Waiting to run
Platform Tests / MacOS (DEV=WEBGPU) (push) Waiting to run
Platform Tests / Windows (DEV=CPU:CLANG) (push) Waiting to run
Platform Tests / Windows (DEV=CPU:LLVM) (push) Waiting to run
Platform Tests / Windows (DEV=CPU:X86) (push) Waiting to run
Platform Tests / Windows (DEV=WEBGPU) (push) Waiting to run
Autogen / In-tree Autogen (comgr 2) (push) Successful in 47s
Unit Tests / Torch Backend Tests (push) Failing after 1m3s
Unit Tests / Linters (push) Failing after 1m27s
Unit Tests / Python Backend (push) Failing after 3m8s
Unit Tests / Fuzzing (push) Failing after 21s
Unit Tests / Docs (push) Successful in 3m32s
Unit Tests / Null Tests (push) Successful in 4m8s
Unit Tests / Unit Tests (push) Successful in 3m30s
Unit Tests / CL IMAGE Tests (push) Failing after 44s
Unit Tests / openpilot Compile Tests (push) Failing after 47s
Unit Tests / Test LLM (push) Failing after 13s
Unit Tests / SPEC=2 (1) (push) Successful in 3m53s
Deploy Docs / deploy (push) Successful in 5m1s
Unit Tests / Optimization Tests (push) Failing after 49s
Unit Tests / SPEC=2 (2) (push) Successful in 3m42s
Unit Tests / Linux (DEV=CL) (push) Failing after 38s
Unit Tests / Models (push) Successful in 1m39s
Unit Tests / Linux (DEV=CPU:X86) (push) Failing after 23s
Unit Tests / Linux (DSP) (push) Successful in 1m47s
Unit Tests / hcq2 (push) Failing after 48s
Unit Tests / Linux (am) (push) Failing after 52s
Unit Tests / Linux (DEV=CPU:LLVM) (push) Successful in 3m13s
Unit Tests / Linux (DEV=CPU:LVP) (push) Successful in 3m6s
Unit Tests / AMD ASM IDE (push) Failing after 2m46s
Unit Tests / Linux (DEV=CPU:CLANG) (push) Successful in 4m19s
Unit Tests / ONNX (CPU) Tests (push) Successful in 5m24s
Autogen / In-tree Autogen (push) Successful in 10m40s
Unit Tests / Linux (DEV=WEBGPU) (push) Failing after 5m0s
Unit Tests / Linux (amdllvm gfx1100) (push) Successful in 3m40s
Unit Tests / Linux (amd gfx1100) (push) Successful in 4m13s
Unit Tests / Linux (amdllvm gfx1201) (push) Successful in 3m35s
Unit Tests / Compile-only (DEV=NULL:IR3:a630) (push) Failing after 30s
Unit Tests / Linux (amd gfx1201) (push) Successful in 4m10s
Unit Tests / Compile-only (DEV=NULL:NAK:sm_120) (push) Failing after 28s
Unit Tests / Compile-only (DEV=NULL:QCOMCL:a630) (push) Failing after 40s
Unit Tests / Linux (amdllvm gfx950) (push) Successful in 3m36s
Unit Tests / Linux (ptx) (push) Failing after 3m17s
Unit Tests / Linux (amd gfx950) (push) Successful in 5m25s
Unit Tests / Linux (nv) (push) Failing after 5m17s
* more lil llm improvements
* default float
2026-08-12 20:13:20 -07:00
George Hotz and GitHub
2297118541
lil llm improvements ( #17513 )
2026-08-12 19:29:32 -07:00
George Hotz and GitHub
3686a1758f
mac/rdma imports lazy ( #17496 )
...
* mac/rdma imports lazy
* ish
* fixes
2026-08-11 17:23:11 -07:00
George Hotz and GitHub
4a253db9b4
minor cleanups to improve import speed ( #17495 )
...
* minor cleanups to improve import speed
* dumb
2026-08-11 16:06:25 -07:00
George Hotz and GitHub
e11df72e0f
notes from digitalocean_mi350x ( #17494 )
...
* notes from digitalocean_mi350x
* cleanup
* revert non-doc changes on digitalocean_mi350x branch
2026-08-11 13:22:11 -07:00
George Hotz and GitHub
7edf80a48a
small changes from new rangeify + remove flaky tc tests ( #17486 )
...
* small changes from new rangeify
* remove test/opt/test_tensor_cores.py
2026-08-10 15:58:49 -07:00
George Hotz and GitHub
566f32fe9f
move platform tests to platform.yml ( #17475 )
...
Autogen / In-tree Autogen (comgr 2) (push) Successful in 41s
Unit Tests / Torch Backend Tests (push) Failing after 53s
Unit Tests / Linters (push) Failing after 1m29s
Unit Tests / Python Backend (push) Failing after 3m5s
Unit Tests / Fuzzing (push) Failing after 22s
Unit Tests / Docs (push) Successful in 3m29s
Unit Tests / Null Tests (push) Successful in 4m5s
Unit Tests / openpilot Compile Tests (push) Failing after 47s
Unit Tests / SPEC=2 (1) (push) Successful in 3m55s
Unit Tests / Optimization Tests (push) Failing after 44s
Unit Tests / Test LLM (push) Failing after 12s
Unit Tests / SPEC=2 (2) (push) Successful in 3m40s
Deploy Docs / deploy (push) Successful in 5m10s
Unit Tests / Linux (DEV=CL) (push) Failing after 36s
Unit Tests / Linux (DSP) (push) Successful in 1m48s
Unit Tests / Models (push) Successful in 3m30s
Unit Tests / CL IMAGE Tests (push) Failing after 5m13s
Unit Tests / Unit Tests (push) Successful in 8m1s
Unit Tests / Linux (DEV=CPU:X86) (push) Failing after 24s
Unit Tests / Linux (DEV=CPU:LLVM) (push) Successful in 3m17s
Unit Tests / Linux (DEV=CPU:CLANG) (push) Successful in 4m13s
Unit Tests / ONNX (CPU) Tests (push) Successful in 5m34s
Unit Tests / hcq2 (push) Failing after 49s
Unit Tests / Linux (DEV=CPU:LVP) (push) Successful in 3m7s
Unit Tests / Linux (am) (push) Failing after 53s
Unit Tests / AMD ASM IDE (push) Failing after 2m53s
Autogen / In-tree Autogen (push) Successful in 12m44s
Unit Tests / Linux (amdllvm gfx1100) (push) Successful in 3m57s
Unit Tests / Linux (amdllvm gfx1201) (push) Successful in 3m45s
Unit Tests / Linux (amd gfx1100) (push) Successful in 4m30s
Unit Tests / Linux (DEV=WEBGPU) (push) Failing after 5m17s
Unit Tests / Linux (amd gfx1201) (push) Successful in 4m26s
Unit Tests / Compile-only (ir3) (push) Failing after 31s
Unit Tests / Compile-only (nak) (push) Failing after 38s
Unit Tests / Linux (amdllvm gfx950) (push) Successful in 4m6s
Unit Tests / Linux (amd gfx950) (push) Successful in 5m55s
Unit Tests / Linux (ptx) (push) Failing after 3m18s
Unit Tests / Linux (nv) (push) Failing after 5m18s
Autogen / In-tree Autogen (macos) (push) Canceled after 0s
Benchmarks / Mac pytest (push) Canceled after 0s
Benchmarks / LLM (DEV=AMD) (push) Canceled after 0s
Benchmarks / LLM (DEV=METAL) (push) Canceled after 0s
Benchmarks / LLM (DEV=NV) (push) Canceled after 0s
Benchmarks / HLB-CIFAR10 (DEV=AMD) (push) Canceled after 0s
Benchmarks / HLB-CIFAR10 (DEV=METAL) (push) Canceled after 0s
Benchmarks / HLB-CIFAR10 (DEV=NV) (push) Canceled after 0s
Benchmarks / MLPerf (AMD) (push) Canceled after 0s
Benchmarks / MLPerf (NV) (push) Canceled after 0s
Benchmarks / Stable Diffusion (DEV=AMD) (push) Canceled after 0s
Benchmarks / Stable Diffusion (DEV=METAL) (push) Canceled after 0s
Benchmarks / Stable Diffusion (DEV=NV) (push) Canceled after 0s
Benchmarks / Tests (DEV=AMD) (push) Canceled after 0s
Benchmarks / Tests (DEV=METAL) (push) Canceled after 0s
Benchmarks / Tests (DEV=NV) (push) Canceled after 0s
Benchmarks / UsbGPU Benchmark (push) Canceled after 0s
Benchmarks / comma Benchmark (0.11.2) (push) Canceled after 0s
Benchmarks / comma Benchmark (0.11.0) (push) Canceled after 0s
Benchmarks / DSP Benchmark (push) Canceled after 0s
Benchmarks / UsbGPU Benchmark (comma) (push) Canceled after 0s
Benchmarks / PCI Driver Benchmark (DEV=AMD) (push) Canceled after 0s
Benchmarks / PCI Driver Benchmark (DEV=NV) (push) Canceled after 0s
Benchmarks / LLVM Speed (push) Canceled after 0s
Platform Tests / MacOS (unit) (push) Canceled after 0s
Platform Tests / MacOS (unit, mock) (push) Canceled after 0s
Platform Tests / MacOS (DEV=METAL) (1) (push) Canceled after 0s
Platform Tests / MacOS (DEV=METAL) (2) (push) Canceled after 0s
Platform Tests / MacOS (DEV=CPU:CLANG) (push) Canceled after 0s
Platform Tests / MacOS (DEV=CPU:LLVM) (push) Canceled after 0s
Platform Tests / MacOS (DEV=CPU:LVP) (push) Canceled after 0s
Platform Tests / MacOS (DEV=WEBGPU) (push) Canceled after 0s
Platform Tests / Windows (DEV=CPU:CLANG) (push) Canceled after 0s
Platform Tests / Windows (DEV=CPU:LLVM) (push) Canceled after 0s
Platform Tests / Windows (DEV=CPU:X86) (push) Canceled after 0s
Platform Tests / Windows (DEV=WEBGPU) (push) Canceled after 0s
Platform Tests / Compile-only (QCOM CL) (push) Canceled after 0s
* ci: split mac/windows/qcom-cl tests into platform.yml
Move the 6 jobs that don't run on Linux (4 macos, 1 windows, 1 QCOM CL
compile test on arm) out of test.yml into a separate Platform Tests
workflow so they run (and can be gated/runners-matched) independently.
* ci: gate platform tests to the upstream repo
Skip mac/windows/qcom-cl jobs anywhere but tinygrad/tinygrad, so the
Platform Tests workflow is disabled on the gitea fork (and any fork).
* ci: revert repo gate on platform tests
Job-level if is only evaluated by gitea when a runner with matching
labels fetches the task; with no mac/windows/arm runners the jobs queue
forever. Disable the workflow on the instance instead.
2026-08-09 23:25:18 -07:00
George Hotz and GitHub
d4d537c8ae
add SPEC checking for the kernel graph ( #17432 )
...
Unit Tests / Null Tests (pull_request) Failing after 22s
Unit Tests / Unit Tests (pull_request) Failing after 22s
Unit Tests / Docs (pull_request) Failing after 34s
Unit Tests / Fuzzing (pull_request) Failing after 26s
Unit Tests / Torch Backend Tests (pull_request) Failing after 57s
Unit Tests / CL IMAGE Tests (pull_request) Failing after 42s
Unit Tests / Test LLM (pull_request) Failing after 12s
Unit Tests / openpilot Compile Tests (pull_request) Failing after 48s
Unit Tests / Linters (pull_request) Failing after 1m29s
Unit Tests / Optimization Tests (pull_request) Failing after 52s
Unit Tests / Linux (DEV=CL) (pull_request) Failing after 42s
Unit Tests / Linux (DSP) (pull_request) Successful in 1m46s
Unit Tests / Python Backend (pull_request) Failing after 3m11s
Unit Tests / Linux (DEV=CPU:X86) (pull_request) Failing after 24s
Unit Tests / SPEC=2 (2) (pull_request) Successful in 3m53s
Unit Tests / SPEC=2 (1) (pull_request) Successful in 4m4s
Unit Tests / Linux (am) (pull_request) Canceled after 0s
Unit Tests / Linux (amd gfx1100) (pull_request) Canceled after 0s
Unit Tests / Linux (amdllvm gfx1100) (pull_request) Canceled after 0s
Unit Tests / Linux (amd gfx1201) (pull_request) Canceled after 0s
Unit Tests / Linux (amdllvm gfx1201) (pull_request) Canceled after 0s
Unit Tests / Linux (amd gfx950) (pull_request) Canceled after 0s
Unit Tests / Linux (amdllvm gfx950) (pull_request) Canceled after 0s
Unit Tests / Linux (nv) (pull_request) Canceled after 0s
Unit Tests / Linux (ptx) (pull_request) Canceled after 0s
Unit Tests / MacOS (unit) (pull_request) Canceled after 0s
Unit Tests / MacOS (unit, mock) (pull_request) Canceled after 0s
Unit Tests / MacOS (DEV=METAL) (1) (pull_request) Canceled after 0s
Unit Tests / MacOS (DEV=METAL) (2) (pull_request) Canceled after 0s
Unit Tests / MacOS (DEV=CPU:CLANG) (pull_request) Canceled after 0s
Unit Tests / MacOS (DEV=CPU:LLVM) (pull_request) Canceled after 0s
Unit Tests / MacOS (DEV=CPU:LVP) (pull_request) Canceled after 0s
Unit Tests / MacOS (DEV=WEBGPU) (pull_request) Canceled after 0s
Unit Tests / Windows (DEV=CPU:CLANG) (pull_request) Canceled after 0s
Unit Tests / Windows (DEV=CPU:LLVM) (pull_request) Canceled after 0s
Unit Tests / Windows (DEV=CPU:X86) (pull_request) Canceled after 0s
Unit Tests / Windows (DEV=WEBGPU) (pull_request) Canceled after 0s
Unit Tests / Compile-only (ir3) (pull_request) Canceled after 0s
Unit Tests / Compile-only (nak) (pull_request) Canceled after 0s
Unit Tests / Compile-only (QCOM CL) (pull_request) Canceled after 0s
Check Line Counts / Check PR Branch status (pull_request_target) Canceled after 0s
Check Line Counts / Core Library Line Difference (pull_request_target) Canceled after 0s
Unit Tests / Linux (DEV=CPU:LVP) (pull_request) Canceled after 1m10s
Unit Tests / Models (pull_request) Canceled after 3m4s
Unit Tests / hcq2 (pull_request) Canceled after 17s
Unit Tests / AMD ASM IDE (pull_request) Canceled after 28s
Unit Tests / Linux (DEV=CPU:CLANG) (pull_request) Canceled after 2m35s
Unit Tests / Linux (DEV=WEBGPU) (pull_request) Canceled after 51s
Unit Tests / Linux (DEV=CPU:LLVM) (pull_request) Canceled after 2m17s
Unit Tests / ONNX (CPU) Tests (pull_request) Canceled after 3m46s
* add SPEC checking for the kernel graph
* skip test
* raise there, not None
* handwritten
* issue with unshard
* multi works
* and bitcast
* fix new tests
* fix linear
* remove call index
* fix shrink
* fixes
2026-08-08 10:00:06 -07:00
1858f1fd9a
viz: collapse PROGRAM nodes like CALL (codex) ( #17438 )
...
Co-authored-by: qazal <[email protected]
2026-08-07 11:44:34 +09:00
George Hotz and GitHub
b4372df9c6
revert wrong custom kernel fix ( #17427 )
2026-08-05 18:31:35 -07:00
George Hotz and GitHub
470c032a5e
fix slice + non contig kernels ( #17423 )
...
* movement: resolve negative int slice bounds against symbolic sizes
negative int bounds in a slice against a symbolic dim were passed through
unresolved, giving wrong views. resolve them against the (possibly
symbolic) size, like slice.indices does for int dims
* schedule: realize custom kernel inputs that don't resolve to a buffer state
rangeify assigns ranges backward from consumers and CALL contributes none,
so the subgraph above a custom kernel input gets no ranges unless something
in it is realized, and reduce conversion crashes with a KeyError. realize
call inputs that don't resolve to a buffer state.
only view-only movement ops preserve the underlying buffer: anything
computed (ALU, REDUCE, ...) must be realized even if one of its sources
resolves to a buffer, since the whole subgraph above the call has no
ranges. unwrapping src[0] unconditionally missed const branches hanging
off non-src[0] children and silently resolved REDUCEs to their source
buffer. includes regression tests for pure const, mixed buffer+const, and
view-over-buffer inputs
2026-08-05 16:18:17 -07:00
George Hotz and GitHub
581bfdd94f
merge track_rewrites and profile_matches into rewrite_group [PR] ( #17420 )
...
* merge track_rewrites and profile_matches into rewrite_group
* bug
* flip ctx polarity
2026-08-05 14:41:38 -07:00
George Hotz and GitHub
757a727808
move callify into tensor ( #17416 )
2026-08-05 11:48:56 -07:00
George Hotz and GitHub
2cce85a606
chat: display reasoning_content from streamed responses ( #17414 )
...
* chat: display reasoning_content from streamed responses
The server's StreamRouter emits reasoning_content deltas for think blocks,
but the chat UI was only reading delta.content, silently dropping all
reasoning. Now reasoning is shown in gray (#888 ) and included in the
message history sent back to the server.
* fix
2026-08-05 10:49:24 -07:00
George Hotz and GitHub
46f0003776
more KernelCountException ( #17407 )
2026-08-04 22:55:38 -07:00
George Hotz and GitHub
e1f42681fa
add new schedule tests + format better ( #17402 )
...
* add new schedule tests + format better
* assert_kernel_count
2026-08-04 18:46:38 -07:00
George Hotz and GitHub
6122b3c98f
use check_schedule in tests where possible ( #17400 )
2026-08-04 18:17:29 -07:00
George Hotz and GitHub
f489f4bc36
add test_eye + color INDEX ( #17394 )
2026-08-04 11:08:08 -07:00
George Hotz and GitHub
c21a552f3d
llm: bugfixes + warmup ( #17384 )
2026-08-03 18:23:14 -07:00
George Hotz and GitHub
3cb786f447
llm: update test_llm_server tests ( #17382 )
2026-08-03 16:06:17 -07:00
George Hotz and GitHub
c2625c78cb
scalar ALU index fix + llm: preserve_thinking ( #17381 )
...
* cstyle: scalar ALU index fix, serve: preserve_thinking, test: fix Handler import
- cstyle.py: return scalar directly when ALU buffer has 1 element
- cli.py: add preserve_thinking param to FallbackTemplate.render
- serve.py: pass preserve_thinking=True when rendering chat completions
- test_llm_server.py: fix import to use Handler from llm.serve
* real fix
2026-08-03 15:10:21 -07:00
George Hotz and GitHub
05bc7c6994
fix llm reasoning and Linear import ( #17372 )
2026-08-02 12:31:52 -07:00
George Hotz and GitHub
fb607fb990
faster devectorizer with one line ( #17360 )
2026-08-01 13:06:35 -07:00
George Hotz and GitHub
15c936db01
merge devectorize + indexing ( #17354 )
...
* external benchmark schedule in 5 sec (codex slop)
* prune
* real?
* delete
2026-08-01 11:40:15 -07:00
George Hotz and GitHub
5a1c641f79
more arg -> val ( #17349 )
...
* more arg -> val
* kimi
* more
2026-07-31 22:44:24 -07:00
George Hotz and GitHub
9082ecef5d
use .val to access the value of Ops.CONST ( #17347 )
2026-07-31 19:51:42 -07:00
George Hotz and GitHub
099d69ff7d
ci: split macos unit test into metal and mock runners ( #17346 )
2026-07-31 19:41:18 -07:00
George Hotz and GitHub
a88f832f0c
remove UOp.val ( #17345 )
2026-07-31 18:58:38 -07:00
George Hotz and GitHub
d65ea465ed
cleanup gemm fragment + add store unshard ( #17313 )
...
* cleanup gemm fragment + add store unshard
* multi
* fix
2026-07-30 20:53:28 -07:00
George Hotz and GitHub
138676ab81
improve fragment example + index unshard (kimi) ( #17288 )
...
* fix dtypes in fragment example
* match tilelang
* flip locals
* fix index on unshard
* test fixes
* kimi needs more taste
2026-07-29 15:38:38 -07:00
George Hotz and GitHub
b30c7e00d4
support 2d on UNSHARD (kimi) ( #17285 )
...
* support 2d on UNSHARD
* fixes
* Fix test and spec
* single barrier
* 2d sharding works for devices too
* cleanups
* no _rewrap
2026-07-29 12:01:59 -07:00
George Hotz and GitHub
52c9e5a99e
rename LOOP -> WEAK and STRONGLOOP -> LOOP ( #17283 )
2026-07-29 10:38:36 -07:00
George Hotz and GitHub
bd296a7359
enable alloc_fragment support with UNSHARD (kimi) ( #17272 )
...
* enable alloc_fragment support with UNSHARD (kimi)
* cleaner with implicit barrier
* cleanups
* cleaner
* strongloop
* dcount cleanups
2026-07-29 09:46:45 -07:00
George Hotz and GitHub
527e57300c
fix smu reset for kernel >= 7 ( #17277 )
...
* fix smu reset for kernel >7
* match kernel
2026-07-28 23:11:21 -07:00
George Hotz and GitHub
451120c6e1
make .barrier implicit (kimi) ( #17275 )
...
* make .barrier implicit (kimi)
* simplier
* lil
* remove tinygrad stock barriers
* readable
* lil
2026-07-28 22:34:57 -07:00
George Hotz and GitHub
dd16d5aead
apply shrink bugfix for 3.11 ( #17271 )
2026-07-28 19:03:06 -07:00
George Hotz and GitHub
a17387d6a0
add UNSHARD to spec ( #17269 )
2026-07-28 17:00:21 -07:00
George Hotz and GitHub
57ae1bc7a7
rename MULTI to UNSHARD ( #17267 )
...
* rename MULTI to UNSHARD
* comment updates (glm)
* rename method to unshard
2026-07-28 16:51:41 -07:00
George Hotz and GitHub
1757067fa6
add device range as src[1] to multi (kimi) ( #17264 )
...
* add device range as src[1] to multi (kimi)
* cleanup
* fix tests
2026-07-28 16:39:34 -07:00
755dfb243b
rename CPU_COUNT to NUM_CPU_THREADS with cgroup awareness ( #17263 )
...
Rename CPU_COUNT to NUM_CPU_THREADS so it can be overridden via env var.
Default uses _get_cpu_count() which respects cgroup limits:
- os.process_cpu_count() on Python 3.13+
- /sys/fs/cgroup/cpu.max on cgroup v2
- /sys/fs/cgroup/cpu/cpu.cfs_quota_us on cgroup v1
- os.sched_getaffinity(0) fallback
Use NUM_CPU_THREADS.value in the dataloader instead of cpu_count(),
and update export_model.py and all renderer references.
Co-authored-by: teeny-runner <runner@teeny>
2026-07-28 15:33:57 -07:00
George Hotz and GitHub
0cdddf3495
remove unneeded default args in renderers ( #17262 )
2026-07-28 14:53:04 -07:00
George Hotz and GitHub
9ce65b71c0
delete dsp_pm_late ( #17260 )
2026-07-28 14:05:41 -07:00
George Hotz and GitHub
7d48926293
switch _device_num to AxisType.DEVICE range (kimi) ( #17252 )
...
* switch _device_num to AxisType.DEVICE range (kimi)
* improve
* don't exclude device for ranges
* current multi ends the range
2026-07-28 13:52:47 -07:00
George Hotz and GitHub
a9ad08064e
make the github actions runners generic for gitea ( #17254 )
2026-07-28 10:48:49 -07:00
George Hotz and GitHub
960430a5e5
Revert "nv: set lower interleave level to reduce GPU hogging (ai slop) ( #15518 )" ( #17218 )
...
This reverts commit ac12914506
2026-07-26 08:15:16 -07:00
George Hotz and GitHub
a346e2ea7b
improve call for viz ( #17187 )
2026-07-24 15:26:05 -07:00
George Hotz and GitHub
f65001e297
cleanups because copy is not allowed in rangeify (kimi) ( #17186 )
...
* cleanups because copy is not allowed in rangeify (kimi)
* gpt cleanups
2026-07-24 14:17:10 -07:00
George Hotz and GitHub
dcad119415
convert COPY -> STORE early ( #17172 )
...
* convert COPY -> STORE early
* finish it (glm)
* no slop
* fixes
* not None
* copy one
* bugs fixed, just kernel count i think
* fix disk to disk
* move out of realize
* revert that
* simpler
* fixes
* add copy kernel simplification
* fixes
* dead copy
* remove reshape on store
* fix mstack lowering
2026-07-24 13:49:17 -07:00
George Hotz and GitHub
08eceb0ad9
llm: fix generated tokens in usage accounting ( #17174 )
2026-07-23 21:38:07 -07:00
George Hotz and GitHub
bf5989ea48
support loops in nir (kimi) ( #17166 )
...
* support loops in nir
* tpying
2026-07-23 17:05:15 -07:00
George Hotz and GitHub
39924387b1
LOOP is srcless RANGE (kimi) ( #17129 )
...
* LOOP is srcless RANGE (kimi)
* upd
* cleanups
* cleanups
* we don't float anymore
2026-07-21 18:41:31 -07:00
George Hotz and GitHub
2a81616492
update rules for INDEX mops ( #17105 )
2026-07-20 18:43:00 -07:00
George Hotz and GitHub
636a43722d
add END and GROUP to addrspace ( #17102 )
2026-07-20 17:26:02 -07:00
George Hotz and GitHub
f7ce7f330d
llm: minor fixes + tests ( #17099 )
...
* llm: minor fixes + tests
* error
2026-07-20 14:31:20 -07:00
George Hotz and GitHub
b53cd35cff
llm: make tokenizer fast (kimi) ( #17077 )
...
* llm: make tokenizer fast
* simpler
* re.escape + qcom mypy fix
2026-07-18 17:31:59 -07:00
George Hotz and GitHub
86a6ad8ed2
llm: split cli.py into serve.py with the HTTP server ( #17065 )
...
* llm: split cli.py into serve.py with the HTTP server
* min edit
2026-07-17 10:45:37 -07:00
George Hotz and GitHub
3ee2baf71d
llm: add tool calling support (kimi) ( #17061 )
...
* llm: add tool calling support
* simpler
* cls
* gpt cleanup
* more gpt cleanups
* tests for tools calling
2026-07-17 10:20:01 -07:00
George Hotz and GitHub
46172bb7c7
llm: add optional jinja template support (kimi) ( #17058 )
...
* add jinja template support (kimi)
* fix tests
* lil
* more crap to fallback
2026-07-16 19:02:58 -07:00
George Hotz and GitHub
d8b83daac6
set tc_upcast_axes to None when done with it ( #17053 )
...
* set tc_upcast_axes to None when done with it
* no tag needed
2026-07-16 14:15:21 -07:00
George Hotz and GitHub
4d8c3d3fc9
add test_hgemm to test_tiny ( #17050 )
...
* add test_hgemm to test_tiny
* dsp skip
2026-07-16 13:12:10 -07:00
George Hotz and GitHub
2b1146b3f4
further clean up wmma ( #17048 )
...
* further clean up wmma
* comment
2026-07-16 11:43:23 -07:00
George Hotz and GitHub
61e104bdfb
use UOp.wmma everywhere ( #17045 )
...
* use UOp.wmma everywhere
* fix
2026-07-16 10:40:48 -07:00
George Hotz and GitHub
8b0dd870ce
use wmma helper ( #17038 )
2026-07-15 23:25:17 -07:00
George Hotz and GitHub
dde2e736e5
fix disable_gc decorator reentrancy ( #16999 )
2026-07-12 15:32:54 -07:00
George Hotz and GitHub
03ecad9486
full removal of dtype.vec ( #16996 )
...
* full removal of dtype.vec
* fix typo
2026-07-12 09:25:50 -07:00
George Hotz and GitHub
246eb51523
more dtype.count removal from x86 and others ( #16989 )
...
* switch x86 to use numel instead of dtype.count
* works?
* remove dtype.count from x86 and others
2026-07-12 09:10:02 -07:00
George Hotz and GitHub
e69ce4be7f
switch x86 to use numel instead of dtype.count ( #16985 )
...
* switch x86 to use numel instead of dtype.count
* works?
2026-07-11 15:26:57 -07:00
George Hotz and GitHub
afeb5c708f
x86 simplification ( #16983 )
...
* simplify x86
* more extras
* simpler
* work
* fixes
* should pasS
* cmt-n
* delete more
* and more
2026-07-11 08:11:05 -07:00
George Hotz and GitHub
43ad225d36
nv_610 support (glm) ( #16979 )
...
* nv_610 support
* unbump onnx
* fix autogen workflow
2026-07-10 20:37:30 -07:00
George Hotz and GitHub
40de90ab19
lil changes to cifar ( #16972 )
...
* lil changes to cifar
* lil changes
* pool
2026-07-10 14:19:46 -07:00
George Hotz and GitHub
77823056d4
remove dead DSP code and pre_matcher ( #16968 )
2026-07-10 11:27:50 -07:00
George Hotz and GitHub
1e55cef493
remove vec from llvm ( #16966 )
...
* remove vec from llvm
* scalars are junk
* remove vec from dsp
2026-07-10 10:47:31 -07:00
George Hotz and GitHub
34452efa22
remove dtypes.vec from cstyle ( #16961 )
...
* remove dtypes.vec from cstyle
* bugfix from gpt5.6
* cleanups
2026-07-09 23:11:11 -07:00
George Hotz and GitHub
d80c971ea2
param + cast dtype in arg ( #16955 )
...
* param + cast dtype in arg
* require dtype
* fix replace in dtype decomps
* fix x86
* dtype none is fine there
* Revert "dtype none is fine there"
This reverts commit 5315cc3e59
2026-07-09 16:32:01 -07:00
George Hotz and GitHub
95681f17ee
remove extra dtypes from tests ( #16954 )
...
* remove extra dtypes from tests
* a few more
2026-07-09 14:59:56 -07:00
George Hotz and GitHub
1df49a7bf5
dtype doesn't need to be explicit in most places now (glm) [PR] ( #16951 )
...
* dtype doesn't need to be explicit in most places now (glm) [PR]
* fix amd pcode
* image idx long/int meh
* real image fix
2026-07-09 14:41:34 -07:00
George Hotz and GitHub
8d4c9d1058
make the upat compiler match the spec (glm) [PR] ( #16948 )
...
* make the upat compiler match the spec
* cleanups
* CUSTOMI/CUSTOM
* fix that test
* pyliteral
* fix variable shadowing
2026-07-09 12:30:39 -07:00
George Hotz and GitHub
904b51a783
dtype fixups (grok) [PR] ( #16940 )
...
* dtype fixups
* a few more
* amd fixes
* revert that
* Remove comment in test_graph_rewrite_div_folding_bug
Removed comment about STACK dtype matching its element dtype in the test for graph rewrite division folding bug.
2026-07-09 11:03:22 -07:00
George Hotz and GitHub
00d01d978d
add dtype production rule checked with SPEC=2 ( #16939 )
...
* add dtype production rule checked with SPEC=2
* no None support yet
* detach passes through
* work
* fix
* line len
2026-07-08 18:04:34 -07:00
George Hotz and GitHub
7f8bbe5407
hlb_cifar free speed ( #16938 )
...
* free speed for hlb cifar
* mirror
2026-07-08 15:49:48 -07:00
George Hotz and GitHub
fdffc6c0c8
remove dtypes base ( #16931 )
...
* remove dtypes base
* find/replace bug
* vcount is junk too
2026-07-08 11:48:11 -07:00
George Hotz and GitHub
63cb1369cb
remove ImageDType (glm) ( #16918 )
...
* remove ImageDType (glm)
* gpt work
* better
* is_image_shape
* cleanups
* remove crap from index
* revert crap python change
* image index with float, this needs to be fixed
* simp
2026-07-08 11:27:52 -07:00
George Hotz and GitHub
6358f939e7
param arg is always the shape ( #16914 )
...
* param arg is always the shape
* earlier
* pm_no_weakints
* real fix
2026-07-08 10:51:56 -07:00
George Hotz and GitHub
6e05cbdbcb
remove the dtype.scalar calls everywhere [pr] ( #16930 )
...
* remove the dtype.scalar calls everywhere
* remove base everywhere
* Revert "remove base everywhere"
This reverts commit 0dd0a5a243
2026-07-08 10:10:03 -07:00
George Hotz and GitHub
e4d0d634d4
remove a bunch of dtype.count [PR] ( #16929 )
...
* remove a bunch of dtype.count
* remove nonsense rules
* rm dead code
2026-07-08 09:34:46 -07:00
George Hotz and GitHub
a951650865
lil cleanups from reduce and expand [PR] ( #16928 )
2026-07-08 08:54:19 -07:00
George Hotz and GitHub
4a51047146
remove unused pm_syntactic_sugar ( #16920 )
...
* remove unused pm_syntactic_sugar
* merge in group for reduce
* merge in add loads
* that ctx is unused
* late loads
* Revert "late loads"
This reverts commit ee70cffcee
2026-07-07 15:40:52 -07:00
George Hotz and GitHub
2aebb6f6c4
move mop cleanup [pr] ( #16915 )
...
* move mop cleanup [pr]
* those are mop cleanups
* more mop_cleanups
* revert
* movement.py
2026-07-07 14:30:32 -07:00
George Hotz and GitHub
2fc7e5341b
final removal of PtrDType (glm) ( #16913 )
...
* final removal of PtrDType (glm)
* junk
2026-07-07 13:37:33 -07:00
George Hotz and GitHub
d8fbbff260
EXPAND adds dims to the front (glm) ( #16908 )
...
* EXPAND adds dims to the front (glm)
* more read images
* simpler + spec
* spec should use n instead of s'
* bump comma to 12
2026-07-07 11:03:06 -07:00
George Hotz and GitHub
0035bb6fa8
only front reduce ( #16901 )
...
* only front reduce
* lint + spec
* that should fail
* meh
2026-07-07 09:55:07 -07:00
George Hotz and GitHub
6f1a983493
Ops.GROUP has no shape, like sink ( #16906 )
...
* Ops.GROUP has no shape, like sink
* readme
2026-07-07 08:20:38 -07:00
George Hotz and GitHub
6e44176cfe
fix flash attention example ( #16900 )
...
* fix spec for amd_copy_matmul
* remove shaped wmma
* fix flash attention example?
* dead code
2026-07-06 20:51:32 -07:00
George Hotz and GitHub
d94ad4444e
fix spec for amd_copy_matmul + remove SHAPED_WMMA ( #16897 )
...
* fix spec for amd_copy_matmul
* remove shaped wmma
2026-07-06 19:00:10 -07:00
George Hotz and GitHub
40d112d4d6
remove ptr=True from index ( #16898 )
2026-07-06 18:34:18 -07:00
George Hotz and GitHub
6b8b2f5aeb
remove PtrDtype from tests ( #16896 )
...
* remove PtrDtype from tests
* fixes
* not needed
* no -1
2026-07-06 18:22:22 -07:00
George Hotz and GitHub
4faed79216
remove placeholder from is_ptr ( #16893 )
...
* remove placeholder from is_ptr
* simpler
* needed
* remove rewriter
* fix tests
2026-07-06 16:55:56 -07:00
George Hotz and GitHub
ccd3428aad
remove ptrdtype from kernel graph ( #16892 )
...
* remove ptrdtype
* tag instead of set
* no ptrdtype
* debuf like the ranges
2026-07-06 14:59:07 -07:00
George Hotz and GitHub
9512dc30f4
remove vec from const ( #16889 )
...
* remove vec from const
* reject in spec
* remove all vector dtypes
* not needed
* remove that
* clean up invalid
2026-07-06 12:23:07 -07:00
