github/tinygrad
1
1
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-01-29 08:48:15 -05:00
Code Issues Packages Projects Releases Wiki Activity
10,417 Commits 209 Branches 14 Tags
reprocess_node_experiments
Commit Graph

10417 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
qazal
ff6def9ffb simple contiguous_while_contiguous prereqs [pr] (#8038) 
* simple contiguous_while_contiguous prereqs [pr]

* early realize

* fine if it's folding a non-contig buffer
 2024-12-04 23:00:28 +08:00
Ahmed Harmouche
c9e7701417 Fast YoloV8 on WebGPU (#8036) 
* Fast yolov8 with downscaled input

* Faster + FPS meter

* Add loader while model is downloading/compiling

* Title touchup
 2024-12-04 15:23:09 +01:00
qazal
b116e1511d make device on uop optional [pr] (#8034) 2024-12-04 20:18:00 +08:00
Ahmed Harmouche
13eedd373b Run WebGPU tests on ubuntu (#8033) 2024-12-04 12:42:04 +01:00
leopf
fb89971e73 use BufferedReader (#8032) 2024-12-04 19:08:54 +08:00
George Hotz
08657cb7b0 hotfix: bump expectations in speed_v_theoretical 2024-12-04 19:00:33 +08:00
George Hotz
ea65c79ba2 hotfix: don't spam BEAM debug in speed_v_theoretical 2024-12-04 18:47:16 +08:00
George Hotz
09b00b1b04 hotfix: use kernel timings instead of python timings in speed_v_theoretical 2024-12-04 18:36:17 +08:00
George Hotz
8f65c1fafb simpler block reorder function [pr] (#8031) 
* simpler block reorder function [pr]

* simpler

* block_reorder in substitute, so wasteful otherwise

* extend and count

* leave push logic for same order

* sort new ctx

* less loop

* Revert "less loop"

This reverts commit 30249d097a.
 2024-12-04 17:57:35 +08:00
leopf
f0401e14e8 tar_extract with Tensors (#7853) 
* initial

* USTAR, PAX and GNU support + testing

* from_bytes byteorder

* use TarInfo.frombuf

* tensor only usage

* remove contextlib.suppress

* shorter ow,pax

* more tests

* testing length + move tests

* cleanup

* new approach: RawTensorIO

* fix fetch

* enable read test

* cleanup and ignore fix

* fix for python < 3.12

* make it RawIO

* functions

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
Co-authored-by: chenyu <chenyu@fastmail.com>
 2024-12-04 17:03:19 +08:00
George Hotz
1e06aefde7 bunch up ops for lines [pr] (#8030) 2024-12-04 17:03:01 +08:00
uuuvn
e9c5b23ba1 Use MTLCompiler directly (v2) (#7920) 
* Use MTLCompiler directly (v2)

* to_block_literal and REQUEST_TYPE_COMPILE

* Rewrite command encoding

* Revert to_block_literal

* Maybe that's more readable to some people?

* Typo and comment about stdlib caching

* Update ops_metal.py

* Update ops_metal.py

* Update ops_metal.py

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2024-12-04 16:36:48 +08:00
George Hotz
bb98bae751 local reordering in block (#8029) 
* local reordering in block

* load (and parents) is highest priority

* minor loads in order

* comments

* explicit depth

* simpler

* matters less, but store early too
 2024-12-04 15:11:29 +08:00
George Hotz
4cb630ac1c hotfix: early INDEX 2024-12-04 14:47:47 +08:00
George Hotz
fdd1e56827 clean up rewrite logic + merge siblings (#8026) 
* clean up rewrite logic [pr]

* simpler

* merge sibling blocks

* no PR
 2024-12-04 13:26:16 +08:00
chenyu
004b2ecff5 remove lt/gt/le/ge from SimpleMathTrait [pr] (#8027) 
just use the dunder methods
 2024-12-04 00:24:33 -05:00
chenyu
39e0fc05f5 update function to not use gt/lt [pr] (#8025) 
pr does not test this, but it's the same
 2024-12-03 22:39:06 -05:00
chenyu
cfd4d19250 replace .lt in rewrite rules with < [pr] (#8024) 2024-12-03 21:34:47 -05:00
chenyu
0c060fa040 update uop and tests to not use lt/gt/le/ge [pr] (#8023) 
just use dunder methods, eventually remove those from ops
 2024-12-03 21:02:52 -05:00
chenyu
03bf9c2985 unused mul add lt rule [pr] (#8022) 2024-12-03 19:38:34 -05:00
nimlgen
7fda464b08 hcq c-like args state (#8020) 
* hcq c-like args state

* ugh

* Dfix

* rename

* i
 2024-12-03 23:53:35 +03:00
qazal
099364ed32 lazy srcs shape mistmatch assert + fix ASSIGN [pr] (#8014) 
* lazy srcs shape mistmatch assert [pr]

* duplicate assert

* base it later

* keep the assert
 2024-12-03 15:40:37 -05:00
ignaciosica
f14dd1488e reduce on wmma (#8016) 2024-12-03 12:46:28 -05:00
chenyu
dacb1ff38a minor nn cleanups (#8018) 
use more .numel and .ndim
 2024-12-03 12:34:52 -05:00
chenyu
35c30f76f2 minor tweak in ptx asm_for_op [pr] (#8017) 
always compare with dtypes instead of name string
 2024-12-03 12:34:22 -05:00
chenyu
a5af4e5596 clean up wgsl_matcher [pr] (#8015) 
use more UPat syntatic sugar and remove unneeded rules
 2024-12-03 11:55:03 -05:00
Ahmed Harmouche
db330a3110 Remove WebGL (#8012) 2024-12-03 16:02:53 +01:00
chenyu
ef3752625b add test case of realize_size with 0 in shape (#8011) 2024-12-03 09:19:50 -05:00
Ahmed Harmouche
8818046940 YoloV8 on WebGPU (#8007) 
Port YoloV8 to WebGPU
 2024-12-03 15:10:41 +01:00
George Hotz
09eac42fd6 cache indexed uops in st [pr] (#8008) 
* cache indexed uops in st [pr]

* remove arg from range
 2024-12-03 21:27:07 +08:00
Sieds Lykles
e44183647f Improved div folding (#7996) 
* First version of div_mod folding together

* Working version with old div folding behaviour

* Test is fixed

* Fix linting

* Happy mypy

---------

Co-authored-by: chenyu <chenyu@fastmail.com>
 2024-12-03 08:11:25 -05:00
George Hotz
32675a8a77 sacrifice ClangGraph on the altar of lines [pr] (#8009) 2024-12-03 21:11:15 +08:00
qazal
5441127417 assert const folding return shape matches [pr] (#8006) 2024-12-03 19:31:06 +08:00
George Hotz
dddfb494d7 don't mutate the uop/lazybuffer, just the Buffer [pr] (#8000) 
* don't mutate the uop/lazybuffer, just the Buffer [pr]

* fix red test

* try different fix

* that

* that's the right fix

* test for fixed behavior

* bump to 3.12
 2024-12-03 19:03:51 +08:00
qazal
ba1183314a const_like can return a valid [pr] (#8005) 
* const_like can return a valid [pr]

* fixup
 2024-12-03 18:42:12 +08:00
qazal
4e91533419 test: don't ref until schedule (#8004) 2024-12-03 18:06:52 +08:00
George Hotz
b8bf5b2787 minor uop speedups [pr] (#8002) 
* minor uop cleaner [pr]

* free uop creation speed by removing WeakValueDictionary

* a lil faster

* disable that test

* lines

* and it doesn't print non hit patterns
 2024-12-03 17:04:48 +08:00
George Hotz
1028b34a20 add typing to basicblocks (#7999) 2024-12-03 15:05:11 +08:00
George Hotz
0905f87b68 hotfix: print only kernel time 2024-12-03 14:25:08 +08:00
chenyu
17d5719a38 add process replay to webgpu tests (#7998) 2024-12-02 20:27:29 -05:00
chenyu
c7bc75e634 alu(c?t0:f0, c?t1:f1) -> c?alu(t0,t1):alu(f0,f1) (#7900) 
* alu(c?t0:f0, c?t1:f1) -> c?alu(t0,t1):alu(f0,f1)

only do if at least one branch is const, so total alu won't increase

* tests and interesting TODO cases
 2024-12-02 17:19:27 -05:00
chenyu
b91fa24387 script to run regressed sd conv on metal (#7995) 
* script to run regressed sd conv on metal

this and other similar `conv2d + add` kernels contributed to most of the speed regression

* # ruff: noqa: E501
 2024-12-02 15:34:27 -05:00
geohotstan
0a2e10be1d add SELU to Tensor (#7993) 
* add selu

* more clean ups
 2024-12-02 10:04:01 -05:00
Ahmed Harmouche
146e1caea3 Downgrade wgpu to prevent sd segfault (#7969) 2024-12-02 15:48:44 +01:00
wozeparrot
077e7e8ed2 fix: private segment sgpr on gfx103x (#7987) 
Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>
 2024-12-02 20:54:50 +08:00
qazal
bb606e5bcf process replayable ops.py changes from delete_lazy [pr] (#7994) 
* process replayable ops.py changes from delete_lazy [pr]

* hotfix: seed tiny_jit
 2024-12-02 19:38:31 +08:00
George Hotz
0c7477b108 no bool in range [pr] (#7988) 
* no bool in range [pr]

* fix llvm

* add arg to range spec

* fix broken test

* forgot this one

* hotfix: test_tiny jit is a real test
 2024-12-02 19:05:16 +08:00
Ahmed Harmouche
8909dbd82c Remove wgpu specific checks from stable diffusion example (#7991) 2024-12-02 11:31:14 +01:00
qazal
e2916ff210 image dtype fixup refactor for delete_lazy [pr] (#7989) 2024-12-02 18:25:13 +08:00
Ahmed Harmouche
5340d3dedf Merge pull request #7986 from tinygrad/atomics-in-smem-wgpu 
Support packed types in smem on webgpu
 2024-12-02 10:38:19 +01:00