Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
176 commits
Select commit Hold shift + click to select a range
28b98e5
nv: move structs to vram (#16184)
nimlgen May 15, 2026
d54fa86
viz/cli: select all calls in graph by default (#16214)
Qazalin May 15, 2026
c6cf9e8
remove test_svd_nonfull_5_5 (#16217)
chenyuxyz May 15, 2026
07a172d
remove noop requires_grad_ calls (#16213)
chenyuxyz May 15, 2026
d62c1d8
remove Tensor.eye override (#16219)
chenyuxyz May 15, 2026
2570bde
update spec for LOAD (#16221)
geohot May 15, 2026
2549b14
fix caformer onnx run (#16222)
geohot May 15, 2026
2c61f65
cl: device extensions in arch (#16220)
sirhcm May 15, 2026
79c0ae5
metal: arch is GPU family (#16223)
sirhcm May 16, 2026
1596943
llama: fix running flat_llama (#16224)
wozeparrot May 16, 2026
2d48d7a
remove more invalid (#16227)
wozeparrot May 16, 2026
e575f77
move debug prints (#16218)
nimlgen May 16, 2026
ebcb7b7
fp8 gemm tests with scale args (#16231)
Qazalin May 16, 2026
bef5f71
fix nolocals and beam (#16232)
nimlgen May 16, 2026
0ddc50d
do not gate backward on requires_grad (#16230)
chenyuxyz May 16, 2026
d95bf39
fp8 gemm speedup (#16236)
Qazalin May 16, 2026
8631b6f
remove use of requires_grad in test/ (#16237)
chenyuxyz May 16, 2026
dcee90a
remove requires_grad use in extra/examples (#16238)
chenyuxyz May 16, 2026
fcdd1af
remove Tensor.detach override [pr] (#16239)
chenyuxyz May 17, 2026
981c121
remove requires_grad= in tinygrad/ (#16241)
chenyuxyz May 17, 2026
5ae4dbd
make slow tests faster (#16244)
chenyuxyz May 18, 2026
bfb2d1f
Revert "fp8 gemm speedup (#16236)" (#16245)
Qazalin May 18, 2026
db639eb
deviceless const from UOp (#16243)
chenyuxyz May 18, 2026
50481ec
cl: check for cl_khr_fp64 (#16246)
sirhcm May 18, 2026
73e6b49
to and shard is noop for deviceless uop (#16247)
chenyuxyz May 18, 2026
7543440
assign for deviceless const source (#16248)
chenyuxyz May 18, 2026
7515824
ci: actually use clang-20, enable bfloat16 (#16249)
sirhcm May 18, 2026
98b8a2b
llama: use hipcc 7.1 version (#16250)
Qazalin May 18, 2026
d532b4f
multi alu with deviceless const (#16251)
chenyuxyz May 18, 2026
18b102f
llama: also use 7.1 comgr, update startup_walltime.sh (#16253)
Qazalin May 18, 2026
a3d59fa
llama: don't save weight (#16252)
wozeparrot May 19, 2026
c1076ed
Tensor.device and UOp.device can be None (#16255)
chenyuxyz May 19, 2026
e694d7f
more deviceless const prerequisites [pr] (#16256)
chenyuxyz May 19, 2026
3f2d401
all tests pass with NOOPT=1 (#16257)
geohot May 19, 2026
a120709
tighten shape spec for broadcasting (#16206)
geohot May 19, 2026
e4350e7
set hipcc mac docker to 7.1 (#16261)
Qazalin May 19, 2026
b3dcf8f
hcq2: split into schedule/realize (#16216)
nimlgen May 19, 2026
a178301
PYTHONREMU: fix CDNA VOP3 conditional writes (#16258)
sirhcm May 19, 2026
74567c1
fix: pass input device to ONNX helper internal tensors (#16242)
shettysach May 19, 2026
b2e8102
25000 lines for x86 backend
geohot May 19, 2026
aa1e59a
X86 with Ops.INS (#14873)
ttomsa May 19, 2026
890b731
more prerequisuite test changed for deviceless const (#16264)
chenyuxyz May 19, 2026
bb2a51f
fix mypy mockgpu and add tinygrad.renderer.isa to packages (#16265)
sirhcm May 19, 2026
7cdd9cb
PYTHONREMU: V_CVT_PK_BF8_F32 saturation (#16268)
sirhcm May 19, 2026
5551574
Remove Ops.VCONST (#16267)
geohot May 19, 2026
da7414d
fix RUN_PICKLE and test it (#16272)
geohot May 20, 2026
361553c
llama: match flat_llama with model_train (#16269)
wozeparrot May 20, 2026
188d7ec
clone can take device (#16271)
chenyuxyz May 20, 2026
7af7b67
relax policy ASSERT_MIN_STEP_TIME to 3.2 (#16273)
chenyuxyz May 20, 2026
170b857
clean up deviceless const _buffer (#16274)
chenyuxyz May 20, 2026
e1715b3
extent jit const error to deviceless inputs (#16276)
chenyuxyz May 20, 2026
1e0fffe
fused ce llama kernel in UOps (#16263)
Qazalin May 20, 2026
fe2d8d1
filter by base_class in pci_scan_bus on macOS (#16282)
sunkencity999 May 20, 2026
4dbe6a2
remove _force_unique from Tensor init (#16277)
chenyuxyz May 20, 2026
19535df
enable broadcasting in _shape (#16285)
geohot May 20, 2026
a01d591
fix: qlinearconv quant params (#16234)
BraunPhilipp May 20, 2026
a88feef
hcq2: cleanups (#16278)
nimlgen May 20, 2026
825f30b
llama: apply_grad saves memory (#16275)
wozeparrot May 20, 2026
58d58c1
remove DEVECTORIZE (#16290)
geohot May 20, 2026
a19fa29
fix x86 nondeterminism (#16293)
Qazalin May 20, 2026
beea463
UOp.clone [pr] (#16295)
chenyuxyz May 20, 2026
da07b28
am: override smu 13_0_7 to 13_0_0 (#16292)
sirhcm May 20, 2026
9e88b08
x86: don't use id (#16296)
Qazalin May 20, 2026
d548f8d
use clone instead of unique_const in allreduce [pr] (#16297)
chenyuxyz May 20, 2026
172f949
move is_dtype_supported to renderer (#16226)
sirhcm May 21, 2026
ec54725
don't use dtype vec for image idx (#16298)
geohot May 21, 2026
dd279ee
print dtype decomp warning in DEBUG=2 (#16300)
sirhcm May 21, 2026
e33e058
set SPLIT_W13=0 for 8b DP by default (#16302)
Qazalin May 21, 2026
a321700
hcq2: multi prereqs (#16304)
nimlgen May 21, 2026
afc5bfa
llama: remove fused grad accum (#16301)
wozeparrot May 21, 2026
6815f28
dtype.vec shapes (#16287)
geohot May 21, 2026
73ea36f
full(buffer=True) (#16311)
chenyuxyz May 21, 2026
fb718a5
llama: realize amax (#16308)
wozeparrot May 21, 2026
0c41317
llama: update 405b scripts (#16309)
wozeparrot May 21, 2026
720a27b
remove many requires_grad= args (#16321)
chenyuxyz May 21, 2026
518e605
only load tinymesa_cpu when LVP is explicitly requested (#16320)
sirhcm May 21, 2026
31424cd
Tensor.requires_grad -> is_param (#16325)
chenyuxyz May 21, 2026
150a82d
start cleaning up dtype tests (#16324)
sirhcm May 22, 2026
9744d51
use more non-buffered const (#16327)
chenyuxyz May 22, 2026
c2d0657
remove getenv(CI) from core tinygrad (#16326)
sirhcm May 22, 2026
3115952
more unique const removal prerequisite (#16328)
chenyuxyz May 22, 2026
bbfe4f8
quantize_fp8 kernels in uops (#16288)
Qazalin May 22, 2026
5d77a94
am: mec_pipe0_reset on gfx12 only (#16331)
nimlgen May 22, 2026
90dbb45
nv: fix boot mem (#16332)
nimlgen May 22, 2026
eeadf26
Fix no module named error (#16305)
googlefan256 May 22, 2026
acc5197
add missing init files, add chat.html to package-data (#16334)
chenyuxyz May 22, 2026
2d48fe8
feat: bump version to 0.13.0 (#16337)
wozeparrot May 22, 2026
26b3b3f
hcq2: move submit lowering to schedule (#16330)
nimlgen May 22, 2026
451f381
start cleanup of the slowest tests (#16339)
sirhcm May 22, 2026
35461d4
ci: cleanup some deps [pr] (#16340)
sirhcm May 22, 2026
149a87d
deviceless const cleanups (#16341)
chenyuxyz May 23, 2026
926d125
update test_stack (#16345)
chenyuxyz May 23, 2026
a891727
hcq2: multi (#16347)
nimlgen May 24, 2026
5d5e028
remove Tensor.from_uop (#16344)
chenyuxyz May 24, 2026
64095cf
use get_buf in exec_kernel (#16356)
nimlgen May 25, 2026
eecd470
fix mailbox comment, add types (#16360)
Qazalin May 25, 2026
68d2102
llama: offload master weights (#16355)
wozeparrot May 25, 2026
2ab90f3
use windows-specific alias nvcuda when loading cuda on windows (#16260)
IntendedConsequence May 25, 2026
b73d2d1
viz/cli: add --interval (#16363)
Qazalin May 25, 2026
cdeb861
invalids is empty [pr] (#16353)
chenyuxyz May 25, 2026
3506eb0
viz: sidebar toggles always recenter (#16366)
Qazalin May 25, 2026
4bcc53e
viz: stable node position for +- toggle (#16367)
Qazalin May 25, 2026
d8f86be
webgpu: shader-f16 support in arch (#16370)
sirhcm May 25, 2026
689ab6a
move buffer view offset to src (#16364)
geohot May 26, 2026
695a006
hotfix: bump Mac pytest timeout to 4 minutes
geohot May 26, 2026
8ddd132
remove getenv(CI) (#16365)
sirhcm May 26, 2026
942cb42
Revert "hotfix: bump Mac pytest timeout to 4 minutes"
geohot May 26, 2026
76fc39c
gather to single device (#16354)
wozeparrot May 26, 2026
41ee7da
script to generate testsig for DSP (#16371)
geohot May 26, 2026
322693d
hotfix: bump Mac pytest timeout to 4 minutes (try 2)
geohot May 26, 2026
032905d
hcq2: simpler (#16361)
nimlgen May 26, 2026
846a809
viz: add +- toggle for hidden UOps (#16368)
Qazalin May 26, 2026
7f1b028
bufferview offset is units of input dtype (#16378)
geohot May 26, 2026
09019d6
remove unique_const (#16372)
chenyuxyz May 26, 2026
9b00def
Revert "remove unique_const (#16372)" (#16380)
chenyuxyz May 26, 2026
bac82d4
fix emu bug in gfx950 (#16381)
geohot May 26, 2026
d861c50
remove unique_const (#16382)
chenyuxyz May 26, 2026
0b88827
remove CONST(UNIQUE) (#16383)
chenyuxyz May 26, 2026
fd96303
remove allow_any_len from store (#16385)
geohot May 26, 2026
d23659d
cleanup some old test skips (#16384)
sirhcm May 26, 2026
3adf7f5
disable flaky cl test (#16388)
sirhcm May 26, 2026
156a443
rename BUFFER_VIEW to SLICE (#16391)
geohot May 27, 2026
3a7a6da
llama: fakedata uses real vocab size (#16389)
wozeparrot May 27, 2026
fa14cde
test update for arange and eye (#16394)
chenyuxyz May 27, 2026
aacc8ad
ci: use ubuntu 24.04 (#16393)
sirhcm May 27, 2026
945ed4f
revert const unique changes (#16395)
chenyuxyz May 27, 2026
3e80f37
skip test_setitem_fancy_on_unrealized_view (#16400)
chenyuxyz May 27, 2026
6da7855
test_custom_kernel_precompile_multidevice (#16401)
chenyuxyz May 27, 2026
bacabf0
webgpu: fix enums (#16402)
sirhcm May 27, 2026
c33b767
bring back test and torch backend change for unique const (#16403)
chenyuxyz May 27, 2026
0c385e3
hcq2 rewrite (#16375)
nimlgen May 27, 2026
452c7d4
llama: don't allocate grad_xw13 in bf16 (#16359)
Qazalin May 27, 2026
171401e
skip modulo by zero in test_dtype_alu (#16404)
sirhcm May 27, 2026
8ee3a37
shrink/pad use (new_shape, offset) (#16405)
geohot May 27, 2026
dac3743
llama: delayed scaling in optim (#16407)
wozeparrot May 27, 2026
b21afb4
marg line cleanup (#16408)
geohot May 27, 2026
88e88d6
viz: click on +- toggles sources (#16409)
Qazalin May 28, 2026
5ee6b6b
fix slice store to remove the index (#16410)
geohot May 28, 2026
202adc6
viz: make call toggle easier to click on (#16411)
Qazalin May 28, 2026
0ae957b
refactor webgpu (#16406)
sirhcm May 28, 2026
d72d8ee
.const() should not ignore dtype (#16412)
chenyuxyz May 28, 2026
edca5df
flip offset and shape in pad and shrink (#16414)
geohot May 28, 2026
b0e49af
hcq2: new multi (#16413)
nimlgen May 28, 2026
c8af163
disable process replay by default (#16419)
sirhcm May 28, 2026
c9adde7
addrspace property (#16418)
geohot May 28, 2026
c87f343
use namespace runners (#16387)
geohot May 28, 2026
36c8ff7
llama: use old scale for dequant in optim (#16417)
wozeparrot May 28, 2026
7d38edf
ci: dev matrix (#16420)
sirhcm May 29, 2026
1e7f1dc
add ParamArgs [pr] (#16421)
geohot May 29, 2026
a909acb
move llvmspeed to benchmarks (#16422)
sirhcm May 29, 2026
69aa205
rename clangjit to clang (#16423)
sirhcm May 29, 2026
6e0d526
ci: autocancel outdated pr jobs (#16424)
sirhcm May 29, 2026
f86966a
llama: optim amax margin (#16425)
wozeparrot May 29, 2026
814d414
viz: set label offset for asm (#16426)
Qazalin May 29, 2026
54cfb79
viz: addrspace little colored box (#16427)
Qazalin May 29, 2026
81bc00c
do not require clearing method_cache in viz tests (#16428)
Qazalin May 29, 2026
fa666ce
remove dead branch in UOp [PR] (#16429)
chenyuxyz May 29, 2026
2d7e5ba
remove vec= from UPat.cvar [PR] (#16430)
chenyuxyz May 29, 2026
6787de9
llama: fix mp (#16434)
wozeparrot May 29, 2026
e2a0434
full derivation of addrspace (#16433)
geohot May 29, 2026
d69aca4
hcq2: rework pm_bufferize (#16431)
nimlgen May 29, 2026
9506b78
fix viz addrspace (#16437)
geohot May 29, 2026
b728070
limit CONST(UNIQUE) to invalids only (#16432)
chenyuxyz May 29, 2026
434cfa9
ci: no fetch in backend tests (#16438)
sirhcm May 29, 2026
ef50a49
ci: macos dev matrix (#16436)
sirhcm May 29, 2026
8ac62b2
fix AffineGrid fusion (#16439)
chenyuxyz May 29, 2026
d943493
ci: remove duplicate op compile test (#16441)
sirhcm May 29, 2026
c23652e
llama: minimize peak init mem (#16440)
wozeparrot May 30, 2026
c377d01
ci: run dsp on tinygrad[testing] (#16442)
sirhcm May 30, 2026
cf55aaf
python prg is pkl uops (#16443)
geohot May 30, 2026
6795c2d
llama: zero grad this way (#16445)
wozeparrot May 30, 2026
29b47a0
llama: update local amax implementation after ParamArgs change (#16446)
Qazalin May 30, 2026
2c2cb33
fix word wrap (#16450)
nimlgen May 30, 2026
989f713
support negative pads in circular pad mode (#16448)
javierdejesusda May 31, 2026
556defa
minor updates from vec removal (#16456)
geohot May 31, 2026
03dbaa0
Merge branch 'upstream/master' into sync-20260605
Jun 5, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions .github/actions/process-replay/action.yml
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,7 @@ runs:
steps:
- name: Run process replay tests
shell: bash
if: env.CAPTURE_PROCESS_REPLAY == '1'
run: |
export PR_TITLE=$(jq -r .pull_request.title "$GITHUB_EVENT_PATH")
export CURRENT_SHA=${{ github.event.pull_request && github.event.pull_request.head.sha || github.sha }}
Expand Down
28 changes: 15 additions & 13 deletions .github/actions/setup-tinygrad/action.yml
Original file line number Diff line number Diff line change
Expand Up @@ -42,7 +42,7 @@ inputs:
required: false
default: 'false'
mesa:
description: "Install mesa"
description: "Install mesa (true, false, cpu)"
required: false
default: 'false'
tinydreno:
Expand Down Expand Up @@ -228,6 +228,11 @@ runs:

sudo chown -R $USER:$USER /var/cache/apt/archives/

- name: Add clang to PATH (Linux)
if: inputs.llvm == 'true' && runner.os == 'Linux'
shell: bash
run: echo "/usr/lib/llvm-20/bin" >> "$GITHUB_PATH"

# **** AMD ****
- name: Setup AMD (Linux)
if: inputs.amd == 'true' && runner.os == 'Linux'
Expand Down Expand Up @@ -281,7 +286,7 @@ runs:
cache-name: cache-gpuocelot-build-1
with:
path: ${{ github.workspace }}/gpuocelot/ocelot
key: ${{ runner.os }}-gpuocelot-b16039dc940dc6bc4ea0a98380495769ff35ed99-rebuild-${{ env.CACHE_VERSION }}
key: ${{ runner.os }}-gpuocelot-f463259669c69abce7b3a0567b6c284f348d0f32-rebuild-${{ env.CACHE_VERSION }}
- name: Cache gpuocelot
if: inputs.ocelot == 'true' && github.event_name != 'pull_request'
id: cache-build
Expand All @@ -290,14 +295,14 @@ runs:
cache-name: cache-gpuocelot-build-1
with:
path: ${{ github.workspace }}/gpuocelot/ocelot
key: ${{ runner.os }}-gpuocelot-b16039dc940dc6bc4ea0a98380495769ff35ed99-rebuild-${{ env.CACHE_VERSION }}
key: ${{ runner.os }}-gpuocelot-f463259669c69abce7b3a0567b6c284f348d0f32-rebuild-${{ env.CACHE_VERSION }}
- name: Clone/compile gpuocelot
if: inputs.ocelot == 'true' && steps.cache-build-pr.outputs.cache-hit != 'true' && steps.cache-build.outputs.cache-hit != 'true'
shell: bash
run: |
git clone --recurse-submodules https://github.com/gpuocelot/gpuocelot.git ${{ github.workspace }}/gpuocelot
git clone --recurse-submodules https://github.com/tinygrad/gpuocelot.git ${{ github.workspace }}/gpuocelot
cd ${{ github.workspace }}/gpuocelot/ocelot
git checkout b16039dc940dc6bc4ea0a98380495769ff35ed99
git checkout f463259669c69abce7b3a0567b6c284f348d0f32
mkdir build
cd build

Expand All @@ -306,10 +311,7 @@ runs:
sudo xcode-select -s /Applications/Xcode_16.2.app/Contents/Developer
CMAKE_ARGS="$CMAKE_ARGS -DBoost_INCLUDE_DIR=$(brew --prefix boost)/include -DBoost_LIBRARY_DIR=$(brew --prefix boost)/lib"
else
curl -fL https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvcc/linux-x86_64/cuda_nvcc-linux-x86_64-11.5.119-archive.tar.xz \
| sudo tar -xJ -C /usr/ --strip-components=1
curl -fL https://developer.download.nvidia.com/compute/cuda/redist/cuda_cudart/linux-x86_64/cuda_cudart-linux-x86_64-11.5.117-archive.tar.xz \
| sudo tar -xJ -C /usr/ --strip-components=1
CMAKE_ARGS="$CMAKE_ARGS -DLLVM_DIR=$(llvm-config-15 --cmakedir)"
fi

cmake .. $CMAKE_ARGS
Expand Down Expand Up @@ -345,13 +347,13 @@ runs:

# **** mesa ****
- name: Install mesa (linux)
if: inputs.mesa == 'true' && runner.os == 'Linux'
if: inputs.mesa != 'false' && runner.os == 'Linux'
shell: bash
run: sudo curl -fL https://github.com/sirhcm/tinymesa/releases/download/v1/libtinymesa_cpu-mesa-25.2.7-linux-amd64.so -o /usr/lib/libtinymesa_cpu.so
run: sudo curl -fL https://github.com/sirhcm/tinymesa/releases/download/v1/libtinymesa${{ inputs.mesa == 'cpu' && '_cpu' || '' }}-mesa-25.2.7-linux-amd64.so -o /usr/lib/libtinymesa${{ inputs.mesa == 'cpu' && '_cpu' || '' }}.so
- name: Install mesa (macOS)
if: inputs.mesa == 'true' && runner.os == 'macOS'
if: inputs.mesa != 'false' && runner.os == 'macOS'
shell: bash
run: brew install sirhcm/tinymesa/tinymesa_cpu
run: brew install sirhcm/tinymesa/tinymesa${{ inputs.mesa == 'cpu' && '_cpu' || '' }}

# *** tinydreno ***
- name: Install tinydreno (linux)
Expand Down
19 changes: 16 additions & 3 deletions .github/workflows/benchmark.yml
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,7 @@ jobs:
CI: ""
CAPTURE_PROCESS_REPLAY: "0"
runs-on: [self-hosted, macOS]
timeout-minutes: 3
timeout-minutes: 4
defaults:
run:
shell: bash -e -o pipefail {0}
Expand Down Expand Up @@ -625,15 +625,15 @@ jobs:
- name: IR3 openpilot compile3 0.11.0 driving_vision
run: BENCHMARK_LOG=ir3_openpilot_0_11_0_vision PYTHONPATH="." ASSERT_MIN_STEP_TIME=17 DEV=QCOM:IR3 FLOAT16=1 IMAGE=1 NOLOCALS=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/v0.11.0/selfdrive/modeld/models/driving_vision.onnx
- name: openpilot compile3 0.11.0 driving_policy
run: BENCHMARK_LOG=openpilot_0_11_0_policy PYTHONPATH="." ASSERT_MIN_STEP_TIME=3 DEV=QCOM FLOAT16=1 IMAGE=1 NOLOCALS=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/v0.11.0/selfdrive/modeld/models/driving_policy.onnx
run: BENCHMARK_LOG=openpilot_0_11_0_policy PYTHONPATH="." ASSERT_MIN_STEP_TIME=3.2 DEV=QCOM FLOAT16=1 IMAGE=1 NOLOCALS=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/v0.11.0/selfdrive/modeld/models/driving_policy.onnx
- name: openpilot compile3 0.11.0 dmonitoring
run: BENCHMARK_LOG=openpilot_0_11_0_dmonitoring PYTHONPATH="." ASSERT_MIN_STEP_TIME=11 DEV=QCOM FLOAT16=1 IMAGE=1 NOLOCALS=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/v0.11.0/selfdrive/modeld/models/dmonitoring_model.onnx
- name: DEBUG=2 openpilot compile3 0.10.1 driving_vision
run: PYTHONPATH="." DEBUG=2 DEV=QCOM FLOAT16=1 IMAGE=1 NOLOCALS=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/720392c9a5b986981fdbed1bb8c47a6c5573a50e/selfdrive/modeld/models/driving_vision.onnx
- name: openpilot compile3 0.10.1 driving_vision
run: BENCHMARK_LOG=openpilot_0_10_1_vision PYTHONPATH="." ASSERT_MIN_STEP_TIME=17 DEV=QCOM FLOAT16=1 IMAGE=1 NOLOCALS=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/720392c9a5b986981fdbed1bb8c47a6c5573a50e/selfdrive/modeld/models/driving_vision.onnx
- name: openpilot compile3 0.10.1 driving_policy
run: BENCHMARK_LOG=openpilot_0_10_1_policy PYTHONPATH="." ASSERT_MIN_STEP_TIME=3 DEV=QCOM FLOAT16=1 IMAGE=1 NOLOCALS=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/720392c9a5b986981fdbed1bb8c47a6c5573a50e/selfdrive/modeld/models/driving_policy.onnx
run: BENCHMARK_LOG=openpilot_0_10_1_policy PYTHONPATH="." ASSERT_MIN_STEP_TIME=3.2 DEV=QCOM FLOAT16=1 IMAGE=1 NOLOCALS=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/720392c9a5b986981fdbed1bb8c47a6c5573a50e/selfdrive/modeld/models/driving_policy.onnx
- name: openpilot compile3 0.10.1 dmonitoring
run: BENCHMARK_LOG=openpilot_0_10_1_dmonitoring PYTHONPATH="." ASSERT_MIN_STEP_TIME=11 DEV=QCOM FLOAT16=1 IMAGE=1 NOLOCALS=1 taskset -c 4-7 python3 examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/720392c9a5b986981fdbed1bb8c47a6c5573a50e/selfdrive/modeld/models/dmonitoring_model.onnx
- name: benchmark MobileNetV2 on DSP
Expand Down Expand Up @@ -806,3 +806,16 @@ jobs:
pkill -f 'extra/remote/serve.py' || true
- name: Run process replay tests
uses: ./.github/actions/process-replay

llvmspeed:
name: LLVM Speed
runs-on: [self-hosted, Linux, tinyboxrandom]
timeout-minutes: 20
if: github.repository_owner == 'tinygrad'
steps:
- name: Checkout Code
uses: actions/checkout@v6
- name: Speed Test
run: DEV=CPU:LLVM THREADS=0 python3 test/speed/external_test_speed_v_torch.py
- name: Speed Test (BEAM=2)
run: BEAM=2 DEV=CPU:LLVM THREADS=0 python3 test/speed/external_test_speed_v_torch.py
Loading
Loading