George Hotz and GitHub
52c9e5a99e
rename LOOP -> WEAK and STRONGLOOP -> LOOP ( #17283 )
2026-07-29 10:38:36 -07:00
George Hotz and GitHub
451120c6e1
make .barrier implicit (kimi) ( #17275 )
...
* make .barrier implicit (kimi)
* simplier
* lil
* remove tinygrad stock barriers
* readable
* lil
2026-07-28 22:34:57 -07:00
George Hotz and GitHub
57ae1bc7a7
rename MULTI to UNSHARD ( #17267 )
...
* rename MULTI to UNSHARD
* comment updates (glm)
* rename method to unshard
2026-07-28 16:51:41 -07:00
wozeparrot and GitHub
c9b60caf8c
gptoss: moe gemm kernels ( #17178 )
2026-07-24 07:25:15 -07:00
qazal and GitHub
abba2aebda
llama: correct fused qkv shape assert ( #17086 )
2026-07-20 15:51:21 +09:00
George Hotz and GitHub
61e104bdfb
use UOp.wmma everywhere ( #17045 )
...
* use UOp.wmma everywhere
* fix
2026-07-16 10:40:48 -07:00
George Hotz and GitHub
8b0dd870ce
use wmma helper ( #17038 )
2026-07-15 23:25:17 -07:00
qazal and GitHub
939f28d571
fused qkv rope custom kernel ( #17021 )
...
* work
* fused qkv_norm
* work
* speed
* not that yet
* test cleanup
* just clone
* remove .realize()
* cleanup tests
2026-07-15 01:08:42 +09:00
qazal and GitHub
62ad646d1c
llama: gemm/fa backward speedups (gpt 5.6) ( #17007 )
...
* fp8 atb gemm speedup
* work
* revert
* fa bw faster
2026-07-14 00:27:00 +09:00
George Hotz and GitHub
03ecad9486
full removal of dtype.vec ( #16996 )
...
* full removal of dtype.vec
* fix typo
2026-07-12 09:25:50 -07:00
qazal and GitHub
9ea7de020b
llama: fuse grad scale in gemm epilogue ( #16992 )
2026-07-12 18:15:01 +09:00
qazal and GitHub
73b38854b4
llama: aT_b fp8 gemm ( #16991 )
2026-07-12 16:57:10 +09:00
chenyu and GitHub
047a467bf9
delete UOp._stack and UOp.vectorize [PR] ( #16988 )
2026-07-11 15:26:55 -04:00
wozeparrot and GitHub
2971343a60
fa: support D=64 and attn sink ( #16935 )
2026-07-08 20:03:47 -07:00
George Hotz and GitHub
fdffc6c0c8
remove dtypes base ( #16931 )
...
* remove dtypes base
* find/replace bug
* vcount is junk too
2026-07-08 11:48:11 -07:00
chenyu and GitHub
e0c70b6b82
mixin/op.py [PR] ( #16926 )
...
* mixin/op.py [PR]
* that
2026-07-07 21:45:52 -04:00
George Hotz and GitHub
2fc7e5341b
final removal of PtrDType (glm) ( #16913 )
...
* final removal of PtrDType (glm)
* junk
2026-07-07 13:37:33 -07:00
George Hotz and GitHub
a145fdce4b
remove the .gep method ( #16867 )
2026-07-04 14:07:09 -07:00
chenyu and GitHub
3a9b87e48f
remove Ops.DEVICE from Ops.PROGRAM [PR] ( #16797 )
2026-06-29 17:41:31 -04:00
wozeparrot and GitHub
bba611bb59
gemm: fix mxfp8 on more shapes ( #16677 )
2026-06-19 13:28:53 -07:00
wozeparrot and GitHub
d37248c3ec
gemm: fix mxfp8 on odd shapes ( #16664 )
2026-06-18 12:03:59 -07:00
wozeparrot and GitHub
bed0c343a3
faster mxfp8 gemm ( #16656 )
2026-06-17 22:35:36 -07:00
qazal and GitHub
f998b9930a
fp8 gemm inv_scale in epilogue ( #16625 )
...
* fuse scale
* remove python inv_scale
* more inv_scale removal
* more cleanups
* cleaner
* diff polish
* work
* rename
* simpler
* simpler
* compute
* c
* Revert "c"
This reverts commit 8941fec7ca .
* Revert "compute"
This reverts commit 9db573a6d3 .
* Revert "simpler"
This reverts commit 910ad33f87 .
* Revert "simpler"
This reverts commit bf75d235a1 .
* s_g
* update types
* less diff noise
* remove
2026-06-15 18:44:41 +09:00
qazal and GitHub
2e77bd01db
fp8 gemm cleanup ( #16607 )
2026-06-13 13:17:32 +09:00
wozeparrot and GitHub
67a4f129c2
llama: fix bf16 gemm oob ( #16603 )
2026-06-12 19:43:05 -07:00
qazal and GitHub
4d34590b7d
llama: less E kernels ( #16517 )
2026-06-12 19:49:25 +09:00
wozeparrot and GitHub
e770805d21
llama: mxfp8 ( #16574 )
2026-06-11 22:15:24 -07:00
wozeparrot and GitHub
2bdc360606
gemm: mxfp8 hipkittens gemm ( #16541 )
...
* gemm: mxfp8 hipkittens gemm
* feat: update hipkittens
* feat: kernel signature
* clean: just kernel
* feat: from tinygrad
* feat: test
* fix: add back utils
* clean: no diff
* clean: no diff
2026-06-09 15:20:05 -07:00
wozeparrot and GitHub
5ef30005fa
update hipkittens ( #16544 )
2026-06-08 18:53:25 -07:00
qazal and GitHub
3b1a5f9770
llama: a_bT and aT_b bf16 gemms ( #16487 )
...
* hk_bf16_gemm
* enable in 8b
* cleanups
* rename to USE_HK_BF16_GEMM
* work
* work
* work
* work
* change the gemms
* work
* work
* set as default
* work
* change
2026-06-04 23:30:21 +09:00
qazal and GitHub
bfb2d1f89a
Revert "fp8 gemm speedup ( #16236 )" ( #16245 )
...
This reverts commit d95bf394e1 .
2026-05-19 02:01:44 +09:00
chenyu and GitHub
dcee90aa3f
remove requires_grad use in extra/examples ( #16238 )
...
except the ones fed into optimizer
2026-05-16 18:40:26 -04:00
qazal and GitHub
d95bf394e1
fp8 gemm speedup ( #16236 )
...
* add asm_gemm option
* milestone
* work
* edit
* only the fast kernel
* diff
2026-05-17 04:58:28 +09:00
wozeparrot and GitHub
528d35e306
llama speed 4 ( #15993 )
2026-04-30 17:14:41 -07:00
chenyu and GitHub
9192c93b7e
Tensor.invalid -> Tesnor.invalids ( #15849 )
...
matches ones and zeros, and to not share name with UOp.invalid
2026-04-21 11:19:51 -04:00
wozeparrot and GitHub
9e60e4a7e7
llama: native fp8 ( #15733 )
2026-04-16 22:16:05 -07:00
wozeparrot and GitHub
457508d5a0
llama: save more 2 ( #15681 )
2026-04-11 01:03:36 -07:00
7e54992bf6
fp8 llama ( #15588 )
...
Co-authored-by: qazal <[email protected] >
2026-04-04 18:24:57 -07:00
645d45d968
DEV has arch ( #15577 )
...
Co-authored-by: Comma Device <[email protected] >
2026-04-03 19:17:19 -04:00
qazal and GitHub
8feb8edc68
gemm/asm: add fp8 support to cdna asm_gemm ( #15542 )
...
* work
* hmm, mixins
* rhs_transposed
* also fix the dtype
* check for hipcc
* Exception
* select dev
* default
2026-03-31 19:32:54 +09:00
George Hotz and GitHub
6e196195d8
add test for flat llama ( #15327 )
...
* add test for flat llama
* simpler
* back to split w1/w3
* env
* still too much ram
* invalid
2026-03-18 15:16:33 +08:00
wozeparrot and GitHub
be23772d43
llama3 fixes part2 ( #15150 )
2026-03-04 23:43:50 -08:00
wozeparrot and GitHub
4e9b85ecfd
fa: pull inputs out of call ( #15127 )
2026-03-04 03:15:49 -08:00
8ebd24637b
fix fa forward building with clang 22 ( #15124 )
...
* fix fa forward building with clang 22
* fix: override rocm path
---------
Co-authored-by: Woze Parrot <[email protected] >
2026-03-04 02:32:25 -08:00
wozeparrot and GitHub
df23057984
fa: change bwd grid dim + unshuffle using mops ( #15068 )
2026-03-04 01:23:40 -08:00
wozeparrot and GitHub
25565b2410
fa: test for mp ( #14907 )
2026-02-22 21:47:36 -08:00
wozeparrot and GitHub
9317e96881
fa: explicitly pass shapes ( #14857 )
2026-02-19 05:26:16 -08:00
wozeparrot and GitHub
45aebe1572
hipkittens fa backward ( #14723 )
2026-02-16 00:38:44 -08:00
George Hotz and GitHub
ac079e43d7
ElementwiseMixin ( #14777 )
2026-02-16 08:50:47 +08:00
33b31d9cd6
tinykittens flash attention dtype fix, add CI ( #14770 )
...
* don't hardcdoe amd device
* add failing tests, ci too
* fix: fix for dtype mixin
* bump to rocm 7.1
---------
Co-authored-by: Woze Parrot <[email protected] >
2026-02-16 01:15:11 +09:00