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
10,931 commits 280 branches 15 tags 325 MiB
Commit graph

273 commits

Author SHA1 Message Date
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
George Hotz
560df206cc
 		split tc test (#12003) 
* split tc test

* split hand coded opts

* remove some skipped tests

* skips on emulated
 2025-09-04 11:47:56 -07:00
George Hotz
9dee724fc4
 		make EMULATE a context var (#12002) 
* make EMULATE a context var

* fix test amx
 2025-09-04 11:15:43 -07:00
nimlgen
897254ad6c
 		ci: add dev<->cpu copy speeds (#11959) 2025-09-02 15:22:44 +03:00
George Hotz
8af8808c61
 		cleanup tests, bump caches (#11746) 2025-08-19 21:21:07 -07:00
George Hotz
1d307f568c
 		move device tests to test/device + test cleanups (#11735) 
* move device tests to test/device

* test speedups

* test device

* linalg to unit

* upd

* so pytest just works

* more divide and skip

* speed

* test devectorize

* add pillow
 2025-08-19 16:02:20 -07:00
wozeparrot
71260a5ea4
 		feat: only bench openpilot 0.9.9 models (#11664) 2025-08-14 19:27:18 -04:00
geohotstan
1e904155e3
 		Add Onnx Huggingface to test/models/test_onnx.py (#11468) 
* BOOM

* cache extra/huggingface/models/

* why max buffer size is not 0

* override MAX_BUFFER_SIZE

* less models

* remove more models and change cache dir to already cached dir

* only metal

* less is more?

* remove check ops

* why is this not setting the ENVVAR

* ughhhhh just test in models

* only cpu and gpu

* only cpu actually

* just override it idk

* final

* move extra dependencies up top

* simplification

* fix print

* make README better

* revert ops_disk fix for now

* clean up test_onnx

* remove testing fashion clip model cuz sloooowwwwww

* actually let METAL run this

* fix comment mistake

* fix download path in run_models

* does this work?

* cleanup setup and teardown

* contextvar like this?

* prove model is cached

* do I need to increment DOWNLOAD_CACHE_VERSION?

* see if cached with incremented DOWNLOAD_CACHE_VERSION

* use warnings to see if the model exists

* revert DOWNLOAD_CACHE_VERSION stuff and clean up

* add retry to download

* nit
 2025-08-14 11:16:41 -04:00
chenyu
b232c60def
 		benchmark openpilot 0.9.9 (#11575) 
* benchmark openpilot 0.9.9

not sure what to do with the 0.9.7 ones with IMAGE=2 and validate

* name
 2025-08-08 01:26:14 -04:00
chenyu
702e38dc19
 		remove FUSE_ARANGE_UINT (#11567) 
also add IGNORE_OOB=1 to bert runs. lowered BS on tinybox to 90 since 96 oom during eval without reset
 2025-08-07 16:49:06 -04:00
chenyu
594cbdc66f
 		skip AM ResNet50 benchmark (#11565) 
hanging with FUSE_ARANGE?
 2025-08-07 14:07:01 -04:00
nimlgen
1afb290027
 		ci: fix runner in nv (#11527) 2025-08-06 10:38:04 +03:00
chenyu
3f742a5a7c
 		comma space lab models benchmark (#11461) 2025-07-31 19:06:18 -04:00
nimlgen
5fc5bb5237
 		ci: clear processes (#11434) 
* unified hcq_smi for managment

* fix

* fix

* no reset for amd
 2025-07-30 22:15:18 +03:00
nimlgen
4b4ba5454c
 		ci: move driver start higher (#11431) 2025-07-30 10:48:38 +03:00
chenyu
204da24cfc
 		increase driverbenchmark timeout-minutes to 15 (#11428) 2025-07-29 19:45:05 -04:00
nimlgen
c88e401d0e
 		ci: fix typos in h machine benchmarks (#11423) 2025-07-29 22:11:47 +03:00
George Hotz
1f1f99c287 hotfix: add DEBUG=3 to driver CI 2025-07-29 11:03:47 -07:00
nimlgen
d38d285489
 		ci: add h machines (#11416) 
* ci: add h machines

* more

* fix names

* names not collide

* 20

* 10
 2025-07-29 19:21:51 +03:00
chenyu
2b48b961be
 		fix a few broken AMX tests (#11204) 2025-07-12 21:42:38 -04:00
George Hotz
0597735f28
 		remove TC=3 not porting this (#11045) 2025-06-30 15:12:49 -07:00