Commit Graph

Select branches

Hide Pull Requests

geohot/updates

gh-pages

master

remove_anchors

#1

#2

#3

v0.10.0

v0.10.1

v0.10.2

v0.10.3

v0.11.0

v0.12.0

v0.13.0

v0.4.0

v0.5.0

v0.6.0

v0.7.0

v0.8.0

v0.9.0

v0.9.1

v0.9.2

Mono Color

• beecac4d85 expand ranges -> unroll outer ranges [pr] (#14440) chenyuandGitHub 2026-01-30 11:26:05 -05:00
• 9eb449f882 clean up toposort sched_sink [pr] (#14439) chenyuandGitHub 2026-01-30 10:18:28 -05:00
• 838cd078bc use atomics for embedding backward (#14400) George HotzandGitHub 2026-01-30 18:10:59 +08:00
• 1998e0bb28 nv: add prof props to dev (#14437) nimlgenandGitHub 2026-01-30 12:51:43 +03:00
• 7a9dee4e50 add call/param UOps (#14433) George HotzandGitHub 2026-01-30 14:51:45 +08:00
• 66d6a68016 viz: sqtt work from cdna gemm (#14434) qazalandGitHub 2026-01-30 00:00:56 -05:00
• 88caf57ef4 ci: unify python versions (#14430) sirhcmandGitHub 2026-01-29 18:42:03 -08:00
• 86a204d22a allow Tensor setitem input to be list/tuple (#14432) chenyuandGitHub 2026-01-29 21:26:58 -05:00
• 4a80319093 clean up split_store final logic [pr] (#14429) chenyuandGitHub 2026-01-29 18:40:07 -05:00
• e47f12f671 ci: replace testing_minimal with testing_unit (#14427) sirhcmandGitHub 2026-01-29 15:02:43 -08:00
• c2fb8b208f fa: 32 block size (#14416) wozeparrotandGitHub 2026-01-29 13:59:13 -08:00
• a979fafae5 cleanup around disk buffer [pr] (#14428) chenyuandGitHub 2026-01-29 16:18:44 -05:00
• dc977a03b0 nv_pma: bw decoder (#14424) nimlgenandGitHub 2026-01-30 00:12:39 +03:00
• ddc041854b failed test case for disk setitem (#14426) chenyuandGitHub 2026-01-29 14:54:19 -05:00
• 31706bf6bc add few more types [pr] (#14425) chenyuandGitHub 2026-01-29 14:04:09 -05:00
• 2d5c24879f nv: pma for 5090 (#14420) nimlgenandGitHub 2026-01-29 20:06:01 +03:00
• c8dc6332d2 memory: read_fields is not universal (#14348) nimlgenandGitHub 2026-01-29 20:00:00 +03:00
• dbe8f034a7 pass z3.Context in validate ctx [pr] (#14423) chenyuandGitHub 2026-01-29 11:11:47 -05:00
• 033ce1b885 types for validate.py (#14422) chenyuandGitHub 2026-01-29 10:56:50 -05:00
• 230d08ec70 test for am recovery and faults handling (#14421) nimlgenandGitHub 2026-01-29 17:11:24 +03:00
• 793afbd473 simplify nn.Embedding, support AFTER in CUSTOM_KERNEL (#14419) George HotzandGitHub 2026-01-29 17:22:13 +08:00
• 0c855d6149 ci: remove unused pydeps (#14418) sirhcmandGitHub 2026-01-28 22:51:26 -08:00
• 4845e42135 llama3 gradacc fixes (#14414) wozeparrotandGitHub 2026-01-28 19:12:39 -08:00
• 37cde4a01a add one line mypy report (#14415) chenyuandGitHub 2026-01-28 20:39:32 -05:00
• 15aed51544 return types for all math.py function (#14413) chenyuandGitHub 2026-01-28 20:10:11 -05:00
• aec1ae0de1 llama: set manual_seed (#14409) nimlgenandGitHub 2026-01-29 01:40:00 +03:00
• 0870ed28b1 add Self type to MathMixin (#14411) chenyuandGitHub 2026-01-28 16:59:38 -05:00
• 079f33c208 fix type in Tensor.mean and Tensor.var (#14410) chenyuandGitHub 2026-01-28 15:24:02 -05:00
• 2b5e99ccc1 minor type cleanups [pr] (#14408) chenyuandGitHub 2026-01-28 14:11:50 -05:00
• 726415dbc8 import sint directly in movement.py TYPE_CHECKING (#14406) chenyuandGitHub 2026-01-28 12:47:26 -05:00
• acb2fc36ba nv_pma: add decoder (#14404) nimlgenandGitHub 2026-01-28 20:44:02 +03:00
• 7b9bc1d8cf _MockMemoryviewMeta for mockgpu (#14405) chenyuandGitHub 2026-01-28 11:59:00 -05:00
• 93793a645b use cl.cl_mem instead of internal ctypes._CData (#14403) chenyuandGitHub 2026-01-28 10:56:41 -05:00
• a9b44070a8 fix webgpu runtime types (#14402) chenyuandGitHub 2026-01-28 10:37:25 -05:00
• 0c6b3f50aa add marker to llama training (#14401) George HotzandGitHub 2026-01-28 22:44:28 +08:00
• 2b7c00d3d2 fix sd-example dtype for CLIP embeddings (#14397) Jakob SachsandGitHub 2026-01-28 15:07:19 +01:00
• a5a9ce3fdf viz: disasm cleanups from null emulate (#14399) qazalandGitHub 2026-01-28 08:03:30 -05:00
• 544928766d hcq_smi: kill mac pids (#14398) nimlgenandGitHub 2026-01-28 15:00:28 +03:00
• 202b74b369 assembly/amd: continue refactors (#14386) George HotzandGitHub 2026-01-28 17:33:03 +08:00
• 5bffa17f82 llama train: better NULL=1 EMULATE=AMD_CDNA4 dev experience (#14395) qazalandGitHub 2026-01-28 03:31:22 -05:00
• 0294014108 fix bufferize cost function for multi, improve VIZ=-1 cli (#14394) qazalandGitHub 2026-01-28 01:53:18 -05:00
• c158acea29 failing multi ram usage test from llama gemm (#14392) qazalandGitHub 2026-01-28 00:32:32 -05:00
• 067e27857e nested composite actions don't work (#14393) sirhcmandGitHub 2026-01-27 21:13:30 -08:00
• 9dddf3d478 don't save caches for PRs, try 2 (#14391) sirhcmandGitHub 2026-01-27 20:30:17 -08:00
• 68fe5d8b36 Revert "don't save caches for PRs (#14389)" (#14390) sirhcmandGitHub 2026-01-27 20:22:26 -08:00
• 4ab228b498 don't save caches for PRs (#14389) sirhcmandGitHub 2026-01-27 20:21:31 -08:00
• 5e36482314 decompose long to ints where unsupported, try 2 (#14383) sirhcmandGitHub 2026-01-27 20:20:43 -08:00
• e496547720 llama3 gradacc (#14291) wozeparrotandGitHub 2026-01-27 19:48:10 -08:00
• 88bc5ee212 assembly/amd: rename to better names (#14384) George HotzandGitHub 2026-01-28 10:00:54 +08:00
• 065b95cfb0 Revert "add retry to fetch (#14370)" (#14385) George HotzandGitHub 2026-01-28 09:35:37 +08:00
• dc4d7f2d55 add retry to fetch (#14370) Eitan TurokandGitHub 2026-01-28 00:04:25 +02:00
• 8d1f3c8885 fix copysign for inf input (#14381) chenyuandGitHub 2026-01-27 16:45:48 -05:00
• 289a3e415e also skip test_nonoverlapping_shrink_assignment (#14382) sirhcmandGitHub 2026-01-27 13:26:26 -08:00
• f34efc1ad1 DISABLE_FAST_IDIV actually works as a ContextVar (#14378) sirhcmandGitHub 2026-01-27 13:12:42 -08:00
• 8c899e4aaf fix copysign for -0 (#14380) chenyuandGitHub 2026-01-27 15:44:58 -05:00
• 62884585a7 failed test case for copysign -0.0 (#14379) chenyuandGitHub 2026-01-27 14:37:17 -05:00
• ec1b28bc2c am: exit early in case of failures (#14376) nimlgenandGitHub 2026-01-27 22:10:02 +03:00
• cd22ee9ed0 add InvalidType to ConstType [pr] (#14373) chenyuandGitHub 2026-01-27 14:09:34 -05:00
• 5b42a1357b SCACHE=0 works with DEBUG (#14377) sirhcmandGitHub 2026-01-27 10:12:43 -08:00
• db010a31be IGNORE_OOB -> CHECK_OOB [pr] (#14374) chenyuandGitHub 2026-01-27 12:20:59 -05:00
• c22667b0c4 also skip test_overlapping_shrink_assignment_reverse (#14375) chenyuandGitHub 2026-01-27 12:20:39 -05:00
• e52d58b041 autogen: update amd (#14372) nimlgenandGitHub 2026-01-27 19:53:14 +03:00
• cbf94a0a95 nv: exit early in case of failures (#14363) nimlgenandGitHub 2026-01-27 19:16:22 +03:00
• ec691cb299 am: print sq intrs (#14366) nimlgenandGitHub 2026-01-27 18:28:13 +03:00
• a5f3d46423 hcq: do not assume kernel names are unique (#14371) qazalandGitHub 2026-01-27 09:03:15 -05:00
• e5df7e640b fix branches in amd_asm_matmul (#14369) George HotzandGitHub 2026-01-27 20:48:42 +08:00
• 0ced258726 HOTFIX: skip crashing assign test geohot 2026-01-27 20:35:17 +08:00
• 131ae604de force_transcendental on sqrt (#14368) George HotzandGitHub 2026-01-27 20:24:41 +08:00
• 14574c68fa Add ContextVar to disable the scheduler cache (#14257) 3 people 2026-01-27 06:55:29 -05:00
• bfc88bcfb8 assembly/amd: emu refactors + enable PYTHON_REMU by default (#14361) George HotzandGitHub 2026-01-27 17:42:24 +08:00
• 2e72625652 Revert "decompose dtypes.long to ints where unsupported (#14261)" (#14362) sirhcmandGitHub 2026-01-26 23:04:59 -08:00
• f866b2a513 mfma loop in asm dsl (#14349) qazalandGitHub 2026-01-26 21:11:37 -05:00
• 0793319929 decompose dtypes.long to ints where unsupported (#14261) sirhcmandGitHub 2026-01-26 15:34:13 -08:00
• a987a4abc3 feat: llama8b dev_beam.sh (#14358) wozeparrotandGitHub 2026-01-26 14:51:23 -08:00
• c9c533fc78 libclang path is homebrew on macos (#14357) sirhcmandGitHub 2026-01-26 14:32:09 -08:00
• d641e63189 improve min/max for AND (#14356) chenyuandGitHub 2026-01-26 15:44:18 -05:00
• f16372487a fix assign hazard on shrink (#14355) chenyuandGitHub 2026-01-26 14:46:30 -05:00
• 145df879c1 find_permutes -> fix_assign_hazard [pr] (#14354) chenyuandGitHub 2026-01-26 14:05:19 -05:00
• e152f1b0f5 llama: use ALL2ALL (#14353) nimlgenandGitHub 2026-01-26 22:01:53 +03:00
• 3f25eb3026 am: ih (#14346) nimlgenandGitHub 2026-01-26 20:11:04 +03:00
• 823bc17fb5 failed test case for shrink overlap assigns (#14350) chenyuandGitHub 2026-01-26 11:58:45 -05:00
• 204f51e739 assembly/amd: bug fixes for PYTHON_REMU (#14347) George HotzandGitHub 2026-01-27 00:48:22 +08:00
• 231305603d remove REAL_DEV [pr] (#14337) chenyuandGitHub 2026-01-26 10:08:16 -05:00
• 9cbe99348a func meshgrid: change param index to type str (#14331) Martin SzewieczekandGitHub 2026-01-26 16:07:56 +01:00
• 3b43d26f10 assembly/amd: emu speed (#14344) George HotzandGitHub 2026-01-26 22:21:34 +08:00
• 774a454bb5 assembly/amd: fix scratch SVE (#14340) George HotzandGitHub 2026-01-26 21:03:51 +08:00
• 2d91fe6310 use amdgpu dsl in mmapeak (#14342) qazalandGitHub 2026-01-26 08:03:43 -05:00
• b2e2ace85b viz: remove ci check, it's VIZ=-1/-2 (#14343) qazalandGitHub 2026-01-26 06:36:23 -05:00
• be23776ba7 assembly/amd: replace pcode with ucode (#14002) George HotzandGitHub 2026-01-26 18:04:29 +08:00
• 984cdc4840 add wrapper class for the -0.0 != 0.0 issue (#14339) George HotzandGitHub 2026-01-26 16:52:37 +08:00
• 92bfe92138 assembly/amd: fix cdna mfma xml (#14329) qazalandGitHub 2026-01-26 03:51:26 -05:00
• 6c109f4d75 LLVM: CPU threading support (#14320) 3 people 2026-01-25 21:12:39 -08:00
• cc49e47ea2 tinygrad changes from ucode (#14336) George HotzandGitHub 2026-01-26 11:30:18 +08:00
• 8477368d07 generic LLVMRenderer class for CPU and AMD (#14321) 3 people 2026-01-25 17:11:49 -08:00
• 11ce1e847d llama train: null device support geohot 2026-01-26 08:53:05 +08:00
• e3601788fa update torch backend function (#14333) chenyuandGitHub 2026-01-25 16:39:34 -05:00
• 9865f51e39 cupti: ref collector (#14330) nimlgenandGitHub 2026-01-25 20:35:21 +03:00
• 21ab23ae18 nv: add pma for ada (#14328) nimlgenandGitHub 2026-01-25 17:33:37 +03:00
• 49db266b96 ReprEnum for repr roundtrips (#14327) George HotzandGitHub 2026-01-25 18:58:31 +08:00
• bf2d9d138f viz: simplify amdgpu cfg (#14326) qazalandGitHub 2026-01-25 01:21:45 -05:00