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
8,229 commits 280 branches 15 tags 325 MiB
Commit graph

8,229 commits

This branch
This branch
All branches
Author SHA1 Message Date
George Hotz
dbb50e4a00 knum 4 2025-03-21 15:48:50 +08:00
George Hotz
71c7c455a6 quantize 2025-03-21 14:55:29 +08:00
George Hotz
ff3438be4e fast 2025-03-21 13:04:18 +08:00
George Hotz
bc5e23061b diasm 2025-03-21 11:22:40 +08:00
George Hotz
5ce951fb34 l2 2025-03-21 11:14:12 +08:00
George Hotz
4a49d05a3f
 		Merge branch 'master' into dsp_search 2025-03-21 10:26:38 +08:00
George Hotz
c3c85c64ee simpler 2025-03-21 09:24:33 +08:00
Sieds Lykles
3ad3ac4d1e
 		Change dtypes.int to dtypes.ints (#9517) 2025-03-20 17:24:26 -04:00
chenyu
b9fab9b914
 		pin ruff to 0.11.0 in CI (#9520) 
0.11.1 had a bug https://github.com/astral-sh/ruff/issues/16874 that breaks ci
 2025-03-20 13:12:50 -04:00
George Hotz
61c02ca634 cleanups 2025-03-20 23:27:06 +08:00
George Hotz
325044bcaf okay that should actually prefetch 2025-03-20 22:59:59 +08:00
George Hotz
91ac508878 prefetch 2025-03-20 22:56:38 +08:00
George Hotz
2ed30f5366 correct flops 2025-03-20 21:46:13 +08:00
George Hotz
d0b9c7e7ca fast like nascar? 2025-03-20 21:27:26 +08:00
George Hotz
f6ed8f4a27 8 folds 2025-03-20 21:20:46 +08:00
George Hotz
87718170d2 more generic 2025-03-20 21:14:33 +08:00
George Hotz
b67af4049c knum 20 2025-03-20 20:59:06 +08:00
George Hotz
16e425a4c0 work 2025-03-20 20:24:21 +08:00
George Hotz
c867a48ab4 custom 2025-03-20 20:02:35 +08:00
George Hotz
2dc82c0604 should be fast 2025-03-20 19:49:04 +08:00
George Hotz
e7402e6643 KNUM=13 will be fast like roadrunner 2025-03-20 18:45:53 +08:00
George Hotz
e5ccd9e846 work 2025-03-20 15:20:03 +08:00
George Hotz
624197f169 swizzle better 2025-03-20 12:41:24 +08:00
George Hotz
d42350a401 simple test 2025-03-20 12:37:29 +08:00
George Hotz
3c5161b4cb
 		add validation of the bounds of Ops.INDEX (#9503) 
* add validation of the bounds of Ops.INDEX

* do mask properly

* more validation

* correct

* fix gated

* add CAST support to vmin/vmax

* fix ptx and image

* ptx no diff

* upat.index also stays

---------

Co-authored-by: qazal <qazal.software@gmail.com>
 2025-03-20 12:15:55 +08:00
qazal
0b20f91ce7
 		remove move_mask from the devectorizer (#9511) 
* remove move_mask from the devectorizer

* add (wrong) ptx

* reason

* enable index addition in PTX, we won't have the INDEX anyways

* space
 2025-03-20 11:53:12 +08:00
qazal
9302738263
 		hotfix: more consistent wgsl.py spacing + cleanups [pr] (#9515) 
* hotfix: more consistent wgsl.py spacing + cleanups [pr]

* free things up
 2025-03-20 11:07:15 +08:00
George Hotz
223feb2118
 		Merge branch 'master' into dsp_search 2025-03-20 10:52:30 +08:00
George Hotz
68053d0510
 		dsp stuff / sniff ioctls from snpe (#9490) 
* sniff ioctls from snpe

* dump input buffers

* snpe logs from dsp

* NHWC support

* knum 3

* this run?

* revert those

---------

Co-authored-by: Comma Device <device@comma.ai>
 2025-03-20 10:38:23 +08:00
qazal
2223b93338
 		add UPat.or_casted [pr] (#9513) 2025-03-20 10:08:32 +08:00
qazal
1839e8c9b3
 		place masks in INDEX for TestGatedStoreRewrite [pr] (#9512) 2025-03-20 09:46:53 +08:00
b1tg
bd731a8624
 		AMDCompiler refactor (no_comgr prereq) (#9497) 
* add amdgpu_disassemble to helpers

* refactor hip compiler

---------

Co-authored-by: b1tg <b1tg@users.noreply.github.com>
 2025-03-20 09:44:07 +08:00
geohotstan
8c0d0a122c
 		Add return_indices to max_pool (#9506) 
* wow argmax is so good

* 1 less line

* clean up and better variable names

* is this torch thing right...?

* add more tests

* slap a TODO on it

* clean ups

* prettier looking code and fix ceil mode test

* add return types and some docs

* ok that was a bad example since indices == value, just no example
 2025-03-19 15:25:37 -04:00
chenyu
189f62d44f
 		add rounding to tqdm unit scale (#9507) 
fixed `AssertionError: ' 1.00/10.0  1000it/s]' != ' 1.00/10.0  1.00kit/s]'`
 2025-03-19 12:08:46 -04:00
nimlgen
a5c971ff3a
 		am: prereqs for rdna4 1/n (#9495) 
* am: ip_ver rename for acc

* am: refactor this

* fix version

* ugh
 2025-03-19 17:14:57 +08:00
Francis Lam
1e5d9ad8f7
 		extra/gemm/max_matmul: start of custom kernels for GEMM (#6926) 
* extra/gemm/max_matmul: start of custom kernels for GEMM

* add an unoptimized FP16/FP16 MMA example

* add slow 3-stage fp16 acc example

* add correct 3-stage pipeline with unswizzled/flat smem input (slow)

* add acc fp16 example with 3 stages and swizzle (no bank conflicts)

* add max version of NV fp16_fp16_fp16

* fix up comments and removed unused code in max variations

* add start of no_xor example

* fix to account for UOps to Ops
 2025-03-19 15:04:57 +08:00
George Hotz
865f23dd7b olmoe memory usage cleanups 2025-03-19 12:28:18 +08:00
b1tg
2c87a22cf2
 		fix prg size calculation when there are adjacent mapped ranges (#9498) 
Co-authored-by: b1tg <b1tg@users.noreply.github.com>
 2025-03-19 11:55:03 +08:00
b1tg
1d71436e6a
 		use libllvm19 in ci (#9494) 
Co-authored-by: b1tg <b1tg@users.noreply.github.com>
 2025-03-19 11:53:32 +08:00
b1tg
a95b489a55
 		nanoGPT train works with tiny torch backend (#9283) 
* train_shakespeare_char.py works

* move aten.where.self_out to tiny_backend_out

* fix memory leak

* corealize in the backward_hook

* Update backend.py

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2025-03-19 11:51:02 +08:00
chenyu
f8976dd2eb
 		enable more webgpu tests (#9502) 
OSX has larger buffer number limit, and it supports fp16 now
 2025-03-18 23:03:54 -04:00
qazal
ae688e4103
 		simple failing test for scheduling parallel reduce [pr] (#9501) 
* simple failing test for scheduling parallel reduce [pr]

* atol
 2025-03-19 10:52:13 +08:00
leopf
e4dad99145
 		nn.state docs cleanup (#8332) 
* doc cleanup

* extension cleanup

* manual definition

* bring back accept_filename for gguf_load

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
Co-authored-by: chenyu <chenyu@fastmail.com>
 2025-03-18 17:16:40 -04:00
chenyu
1ea4876dfa
 		olmoe touchups (#9499) 
GlobalCounters.reset() and only validate if temperature is 0
 2025-03-18 15:25:45 -04:00
geohotstan
f7506c6c25
 		JIT OLMoE (#9396) 
* jit the forward

* might timeout, idk just send it

* this is dumb

* naive bitonic lol

* idk if this is correct, but that squeeze before is definitly not

* vectorized bitonic sort, but still slow

* yay 1 layer is correct

* alright its pretty good

* good enough

* rerun CI

* nit improve comment
 2025-03-18 14:49:02 -04:00
Ignacio Sica
5c56cac0a0
 		MI300 mfma support (#9417) 
* add f16/f32 mfma support for MI300

- add 16x16 mfma shape support for f16 with f32 acc
- add ops_python mfma emulation
- add arch to AMDRenderer

* minor cleanup

* minor cleanup

* add mfma emulation task to ci

* add back todo

* hotfix: comment

* add tc=3 job to ci
 2025-03-18 14:33:30 -03:00
hooved
5500887eed
 		improve reproducibility of WebGPU CI puppeteer test (#9496) 
* try to make CI test fail with slow JS import

* prevent race between model import and reference

* revert artificial delay in JS module import
 2025-03-18 09:27:38 -04:00
qazal
cde4fd3be3
 		do not view_left assign + elementwise sources always have a shape [pr] (#9491) 2025-03-18 17:42:51 +08:00
George Hotz
117b7a16ef
 		VALIDATE_WITH_CPU [pr] (#9488) 
* VALIDATE_WITH_CPU [pr]

* fix test
 2025-03-18 15:15:04 +08:00
qazal
935cd01f56
 		simple failing test for graph_rewrite children [pr] (#9489) 
* simple failing test for graph_rewrite children [pr]

* lint

* update too
 2025-03-18 13:07:21 +08:00