mirrors/tinygrad
2
0
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-06-24 02:14:17 +00:00
Code Issues Projects Releases Packages Wiki Activity
11,549 commits 280 branches 15 tags 325 MiB
Commit graph

293 commits

Author SHA1 Message Date
chenyu
f5090192c8
 		reorder AMD tensor core benchmark test (#13860) 
* reorder AMD tensor core benchmark test

* disable that
 2025-12-28 12:29:51 -05:00
George Hotz
4702da41d5 hotfix: mkdir for extra/disassemblers 2025-12-19 17:18:37 -04:00
Christopher Milan
97103831c5
 		Revert "remove image from BufferSpec (#13636)" (#13761) 
This reverts commit 2571a1eb47.
 2025-12-19 13:54:36 -05:00
Christopher Milan
2571a1eb47
 		remove image from BufferSpec (#13636) 
* remove image from BufferSpec

* cl tiny_gemm (64) works

* mypy

* padding

* openpilot CL

* reshape properly

* remove extra qcom checks

* pad output

* mypy

* update compile test

* move undo

* TestImageCopy valid images

* TestImageRealization valid images

* TestImageDType valid images

* cleanups

* test_renderer_failures

* ruff

* mypy

* simplify ops_qcom

* bump step time
 2025-12-19 13:41:20 -05:00
George Hotz
4b741e893f
 		remove REMOTE=1 (#13722) 
* remove REMOTE=1

* leave ibverbs
 2025-12-16 15:58:10 -04:00
George Hotz
7589c897b2
 		split usbgpu tests into their own benchmark [pr] (#13711) 2025-12-15 21:42:40 -04:00
qazal
6bafd90248
 		remove unused process replay input [pr] (#13712) 2025-12-16 09:29:35 +08:00
nimlgen
cbae33003d
 		ci: add usb4 (#13643) 
* ci: add usb4

* debug=3

* undef

* revert
 2025-12-11 19:41:41 +03:00
chenyu
2471b49e45
 		minor bert / llama change from grad acc branch (#13622) 
* minor bert / llama change from grad acc branch

* revert those
 2025-12-08 16:04:14 -05:00
chenyu
ac1227575f
 		IMAGE=1 driving_vision in benchmark (#13587) 2025-12-05 10:20:54 -05:00
chenyu
8902781dc1
 		enable more benchmarks (#13540) 
* enable more benchmarks

* disable some

* adjust ASSERT_MIN_STEP_TIME

* mac NOCLANG=1
 2025-12-02 20:31:14 -05:00
nimlgen
455dd88236
 		nv: minimal hevc (#13502) 
* nv: minimal hevc

* validate

* not needed

* tralin

* var

* cpu

* fxi

* desc

* move

* cleanup
 2025-11-30 16:46:55 +03:00
wozeparrot
1f648bb1ba
 		feat: reenable mobilenetv2 dsp (#13320) 2025-11-21 15:21:49 -08:00
chenyu
6372c95094
 		disable benchmark MobileNetV2 on DSP (#13305) 
failed on tinyc2
 2025-11-16 09:42:52 -05:00
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
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
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
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
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
George Hotz
5eb87ab131 hotfix: bump cifar time to 350 2025-10-30 17:29:20 +08:00
b1tg
bb307b9e81
 		fix fp8 vectorization (#12977) 
* fix fp8 vectorization

* add fp8 tc to benchmark
 2025-10-28 13:55:30 -04:00
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
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
George Hotz
726988fa4b
 		late ifs try 2 (#12865) 
* late ifs try 2

* fix image

* fix that test

* panic

* ptx fixups

* preserve toposort

* those pass locally

* Revert "those pass locally"

This reverts commit 063409f828.

* no ls

* make that explicit
 2025-10-22 18:49:27 +08:00
chenyu
6d86e962c7
 		update ASSERT_MIN_STEP_TIME (#12857) 
0.10.1 driving_policy is good now, still need driving_vision and dmonitoring to be fast
 2025-10-21 22:46:07 -04:00
wozeparrot
62e7b8b870
 		feat: just use compile3 (#12849) 2025-10-21 07:56:50 -07:00
wozeparrot
990e8b97ee
 		feat: log openpilot 0.10.1 times (#12816) 2025-10-20 18:30:34 -07:00
chenyu
350a4754a9
 		Update openpilot models (#12780) 
* Update openpilot models

* Update slower model

* fix that

---------

Co-authored-by: Bruce Wayne <harald.the.engineer@gmail.com>
 2025-10-18 20:32:35 -04:00
Harald Schäfer
addc54b96c
 		Simplify openpilot compile3.py (#12748) 
* Simpler compile3

* tests

* remove default args

* onnx file is still fp16

* self-test FP16 too

* allow test disable

* absurd tolerance

* Just do latest

* Try simplest

* use later models

* kernel count not relevant if speed is good

* dead improts

* Revert "dead improts"

This reverts commit f68c2cd15d.

* Revert "kernel count not relevant if speed is good"

This reverts commit 0955ca4ee0.

* add back kernal count check on latest model
 2025-10-18 10:12:22 -04:00
chenyu
285534ce64
 		delete DONT_REALIZE_EXPAND and DONT_GROUP_REDUCES (#12744) 
does nothing now
 2025-10-16 14:11:33 -04:00
chenyu
53478c741d
 		relax ASSERT_MIN_STEP_TIME for space lab policy (#12742) 2025-10-16 11:40:36 -04:00
chenyu
b8cf35fb77
 		print macOS version in CI (#12705) 2025-10-15 15:05:33 -04:00
chenyu
89df6f611d
 		reenable sdxl mac benchmark (#12680) 
also updated faster sd step times
 2025-10-14 17:36:17 -04:00
Sieds Lykles
e625c27598
 		update min step times openpilot (#12600) 2025-10-10 11:24:27 +02:00
chenyu
be05028419
 		move ASSERT_MIN_STEP_TIME to compile3 (#12535) 
threshold is current time +20%
 2025-10-08 22:16:59 -04:00
chenyu
5986d656a2
 		tighter ASSERT_MIN_STEP_TIME (#12531) 
set to about 1.2x of actual time now
 2025-10-08 21:22:54 -04:00
George Hotz
3b0b3a2e64
 		fast RANGEIFY (#12504) 
* rtoposort is fast, can replace rangeify with this

* fast rangeify

* work

* fast rangeify works for mnist

* should work

* progress

* pad fix

* FAST

* tests passing

* don't delete those shape ops

* put in rangeify map

* ending ranges fix

* tests

* mstack/mselect no hacks

* move to indexing.py

* touch up tests + add comments

* disable failing test

* actually make the file readable

* failing

* error
 2025-10-08 19:38:06 +08:00
chenyu
eb3bc277b3
 		remove ASSERT_MIN_STEP_TIME in external_benchmark_openpilot (#12495) 
should add for compile3 and compile 3 only
 2025-10-07 22:13:42 -04:00
chenyu
fe774a4319
 		more skip WINO on benchmark (#12482) 2025-10-07 03:43:51 -04:00
chenyu
8ad5f9e74f
 		skip slow benchmarks (#12481) 
* skip slow benchmarks

padded tc is already slow, rest are slow with rangeify (correct if run locally)

* relax more
 2025-10-07 03:28:56 -04:00
Sieds Lykles
e74be4a140
 		UOp.factor and add chain sorting (#12413) 
* add ordering

* fix some tests

* fix more tests

* shorten comment

* update test

* add rule and test

* add rule and test

* remove check

* use fold_divmod_congruence instead of simplify

* adjust tests

* shorten line

* new algo

* add test

* add function to un-nest the div

* add UOp.factor

* test UOp.factor

* uop_given_valid tries to factor simplex expression

* shorten line

* symbolic_flat is back

* change that back

* fix those new tests

* new rule for ordering

* factor multiple factors

* no symbolic_flat

* symbolic_flat to there

* move that back

* fix imports

* merge correctly

* linter happy

* add rule

* add a test

* cleanup

* revert that for now

* UOp.factor returns self instead of None

* try all_candidates

* remove or_else

* post index symbolic

* add test

* maket this closer to the original

* increase mac hlb_cifar min step time

* add some ordering tests

* cleanup

* increase pytest timeout time

* check dtype
 2025-10-04 06:05:38 +02:00
chenyu
494bb12500
 		skip slow cifar bf16 on red benchmark (#12213) 
very slow to compile the fake bf16
 2025-09-16 14:55:01 -04:00
chenyu
419e997187
 		increase benchmark timeout (#12212) 
account for compile cache, and it's annoying that job died due to timeout also messes the machine
 2025-09-16 14:09:02 -04:00
nimlgen
fb96394ff5
 		auto-select available compilers (#12094) 
* device: auto select compilers

* fix

* metal+opencl

* nv/cuda

* test without ptx

* ptx

* fix tests

* fix

* fix test

* rename

* test + cleaner

* xx

* ops

* better test

* win?

* um?

* types

* debug

* win??

* sep rung

* wtf?

* debug

* skip win

* revert this

* types
 2025-09-10 19:52:01 +03:00
Sieds Lykles
5b73076e48
 		assert benchmark times (#12042) 
* assert jitted times in openpilot

* better error

* better error

* add ASSERT_MIN_STEP_TIME to more models

* t is step_times

* update benchmark times

* update times
 2025-09-09 23:40:02 +02:00
nimlgen
1c6c42715f
 		unify cpu and llvm (#11982) 
* try unify cpu and llvm

* fixes

* fix

* ops

* no llvm

* fix

* rm

* lvmm is ot

* oops

* override

* no llvm

* ignore

* skip llvm

* ooops
 2025-09-09 13:54:44 +03:00
George Hotz
433581f8ed
 		make POSTOPT=2 the default (#12034) 
* make POSTOPT=2 the default

* more matching tc

* fix winograd

* fix that test

* add matvec to Scheduler

* flip tc sort order

* similar speed

* fix beam on image

* disable slow tests

* slow
 2025-09-05 14:34:05 -07:00