github/tinygrad
1
1
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-01-10 07:28:15 -05:00
Code Issues Packages Projects Releases Wiki Activity
11,577 Commits 203 Branches 13 Tags
42abb0586c81e83cb44b0ca9269ee86a4de9f3ba
Commit Graph

1161 Commits

Author SHA1 Message Date
Christopher Milan
4043489803 set curl -f in setup-tinygrad (#13389) 
* set curl -f in setup-tinygrad

* test bad redirect

* Revert "test bad redirect"

This reverts commit ad945e7ffc.
 2025-11-20 13:45:47 -05:00
Christopher Milan
0901a40685 Revert "autogen: fix formatting on zero-argument function-like macros (#13386)" (#13387) 
This reverts commit 58d85d4bab.
 2025-11-20 12:45:35 -05:00
Christopher Milan
58d85d4bab autogen: fix formatting on zero-argument function-like macros (#13386) 
* fix formatting on zero-argument function-like macros

* autogen tests should run

* ugh
 2025-11-20 12:11:04 -05:00
Roelof van Dijk
0dc2ff431d fix: revive torch backend (#13280) 
* fix: revive torch backend

* as_strided view vs copy

* Revert "as_strided view vs copy"

This reverts commit 82a61223f2.

* add extra tests (move inplace, add fusion tests)

* better fusion with inplace_op

* no optimizer hooks (break mnist training fusion)

* split off fusion tests in separate file, assert on resnet fusion

fix: remove comments

* cleanup, reduce diff

* reduce diff

* better fusion and identity checks

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2025-11-19 15:26:50 -08:00
George Hotz
1a332afa76 spec test on 3.14 (#12957) 2025-11-19 00:43:04 -08:00
chenyu
6372c95094 disable benchmark MobileNetV2 on DSP (#13305) 
failed on tinyc2
 2025-11-16 09:42:52 -05:00
Christopher Milan
5b823af696 Remove (pypi) clang dep for autogen (#13284) 
* no more clang

* regen comgr_3

* ci doesn't need pypi clang

* fix objc

* REGEN for libclang

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2025-11-15 09:05:11 -08:00
George Hotz
df53c62a9f bump line count 2025-11-15 08:16:20 -08:00
Christopher Milan
d1bb08c5a1 In-tree autogen: objective c (#13223) 
* checkout changes from autogen branch

* move assert

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2025-11-14 14:08:42 -08:00
nimlgen
14eb48b13a autogen: rename nv_gpu to nv_570 (#13273) 
* autogen: rename nv_gpu to nv_570

* rename
 2025-11-14 20:07:19 +08:00
George Hotz
44d84228ff move comgr_3 logic back to the old place (#13266) 
* move comgr_3 logic back to the old place

* explicit
 2025-11-13 20:05:54 -08:00
Christopher Milan
09f3aae169 In-tree autogen: all C libraries (#13220) 
* checkout files from autogen branch

* ioctl with payload

* fix am generations

* properly fix generations

This reverts commit b2a54f4f41.

* revert discovery.h

* support pragma pack(1)

* typo

* better getter

* typo

* NVCEC0_QMDV05_00_RELEASE[01]_ENABLE

* align support

* anon handling fix

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2025-11-13 18:57:44 -08:00
Harald Schäfer
3af231904e openpilot compile tests: assert pre-rangify speeds (#12775) 
* assert pre-rangify speeds

* typo

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2025-11-13 09:39:06 -08:00
George Hotz
263b724143 one cache and bump it (#13258) 2025-11-13 07:33:31 -08:00
chenyu
3f939f3d3c update pm_simplify_valid (#13241) 
* update pm_simplify_valid

fixed openpilot conv regression

* IMAGE training is broken
 2025-11-12 19:40:02 -05:00
George Hotz
ab9fa964d8 DISABLE_COMPILER_CACHE -> CCACHE (#13234) 
* DISABLE_COMPILER_CACHE -> CCACHE

* Fix cachekey assignment in Compiler constructor
 2025-11-12 15:07:09 -08:00
Christopher Milan
41a098a82d In-tree autogen: libc.py (#13217) 
* checkout changes from autogen branch

* parents

* pylint happy

* move sys to system in helpers.py

* typo

* typo
 2025-11-11 19:13:48 -08:00
chenyu
23b90945c3 add a benchmark for openpilot vision with DEBUG=2 (#13219) 
see per kernel speed, also disable the jobs for 0.9.9
 2025-11-11 14:41:52 -05:00
Gaétan Lepage
6fd7ce3832 migrate to pyproject.toml (#13189) 
* migrate to pyproject.toml

* move mypy config to pyproject.toml
 2025-11-11 09:09:27 -08:00
chenyu
60e55d9a2d line count 18500 (#13191) 2025-11-10 13:52:13 -05:00
chenyu
6c48c87e51 improved ASSERT_MIN_STEP_TIME (#13182) 
* improved ASSERT_MIN_STEP_TIME

getting close, current time +1ms  then round up

* relax
 2025-11-09 16:41:12 -05:00
chenyu
e1d46de8f8 update GROUPTOP heuristic more (#13178) 
reverts #13176
 2025-11-09 02:31:12 -05:00
chenyu
8e868dced8 only GROUPTOP one reduce kernel (#13176) 
* only GROUPTOP one reduce kernel

* ALLOWED_GATED_READ_IMAGE=148
 2025-11-08 22:38:44 -05:00
George Hotz
42b34cf83d bottom up linearizer (#13133) 
* bottom up linearizer

* late stores

* more complete

* remove broken heuristic

* upcast size

* opt

* more conservative

* it needs that

* disable opencl half on QCOM

* fix

* make that a real test

* cpu test okay

* ptx skip

* end is after the range
 2025-11-06 15:30:32 -08:00
chenyu
54141e9cb9 DISABLE_COMPILER_CACHE=1 in speed_v_theoretical (#13096) 2025-11-04 11:28:18 -05:00
chenyu
ddf01fdb15 revert mlperf.yml setting (#13080) 2025-11-03 15:24:13 -05:00
chenyu
a317d6e625 extra/amdpci/setup_python_cap.sh (#13070) 2025-11-02 19:19:36 -05:00
chenyu
ad501ce50a mlperf cron install tqdm (#13069) 
one more...
 2025-11-02 18:09:27 -05:00
chenyu
2c8d619147 mlperf cron install influxdb3-python (#13068) 2025-11-02 17:55:40 -05:00
chenyu
4c22f089fc mlperf cron install tensorflow try 2 (#13067) 2025-11-02 17:11:01 -05:00
chenyu
c58cf91850 mlperf cron install tensorflow (#13066) 2025-11-02 16:48:05 -05:00
chenyu
74db65cf72 update mlperf bert LOGMLPERF (#13065) 2025-11-02 15:26:37 -05:00
chenyu
b18293de96 train bert in mlperf cron (#13064) 
more relevant now
 2025-11-02 15:04:02 -05:00
George Hotz
036ee9f84c Self type + mixins (#13056) 
* use Self type

* mixin

* fix later
 2025-11-02 13:30:01 +08:00
George Hotz
65a0a31475 AMD mi350x matmul from stream (#13040) 
* works

* working mfma

* 120 TFLOPS

* regs

* 192 TFLOPS

* try pipelining

* something

* notes

* contract

* linter to 3.11

* that was a bug
 2025-11-01 17:55:19 +08:00
nimlgen
f6786c1bfd autogen: py314 (#13038) 
* autogen: py314

* bump py?
 2025-11-01 04:02:19 +08:00
George Hotz
5eb87ab131 hotfix: bump cifar time to 350 2025-10-30 17:29:20 +08:00
nimlgen
4b001ec723 amd: pmc in mockgpu (#13000) 
* amd: pmc in mockgpu

* fix

* do not open in ci
 2025-10-30 01:52:02 +08:00
b1tg
bb307b9e81 fix fp8 vectorization (#12977) 
* fix fp8 vectorization

* add fp8 tc to benchmark
 2025-10-28 13:55:30 -04:00
George Hotz
5e01cc299b zero len ranges fail (#12974) 
* zero len ranges fail

* fix Python backend

* fix llvm

* fix ptx

* yolo fix nir

* this works...

* always store...

* always store...

* Revert "always store..."

This reverts commit 0816cf344d.
 2025-10-28 22:49:55 +08:00
George Hotz
e936aa7974 cleanups from if range branch (#12973) 2025-10-28 20:58:47 +08:00
George Hotz
2832954bcb test with IGNORE_OOB=0 (#12960) 2025-10-28 10:32:19 +08:00
George Hotz
7784cec48e pytest-split on spec (#12959) 2025-10-28 10:09:01 +08:00
b1tg
45e2f916a3 add quantize fp8 in llama3 (#12893) 
* add quantize fp8 in llama3

* don't truncate fp8 alu result

* cast to float32 before matmul

* --model weights/LLaMA-3/8B-SF-DPO/

---------

Co-authored-by: chenyu <chenyu@fastmail.com>
 2025-10-27 10:22:57 -04:00
George Hotz
25c2da1579 check SPEC=2 in CI (#12945) 
* check SPEC=2 in CI

* split SPEC=2

* fast enough
 2025-10-27 21:53:57 +08:00
George Hotz
8a941d95a4 SPEC=2 is full spec, SPEC=1 is default (#12910) 
* SPEC=1 passes all tests

* just use SPEC, not __debug__
 2025-10-25 11:10:43 +08:00
chenyu
4b7329001d clean up test_avg_pool3d (#12905) 2025-10-24 14:31:36 -04:00
chenyu
154b4f9f40 test FUSE_OPTIM=1 test/test_optim.py (#12895) 2025-10-23 15:54:27 -04:00
wozeparrot
6e00dec95d feat: pin openpilot 0.10.1 models (#12878) 2025-10-22 14:57:54 -07:00
chenyu
f0831c8c30 add 0.10.0 to comma benchmark (#12875) 
* add 0.10.0 to comma benchmark

disabled the 0.10.1 ones which are pinned to master. it does not work because benchmark uses the cached old version

* that's pinned
 2025-10-22 15:18:21 -04:00