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

• 407ca54382 symbolic fold double where (#9436) chenyuandGitHub 2025-04-05 05:12:17 -04:00
• 9c2fc695b5 cond.logical_not().where(a,b) -> cond.where(b,a) (#9741) 3 people 2025-04-05 01:13:32 +02:00
• e9a3ac02a5 Remove ne from arange pattern (#9743) Sieds LyklesandGitHub 2025-04-05 00:31:13 +02:00
• 86c55414d7 ops_amd: simplify gfx version (#9742) nimlgenandGitHub 2025-04-04 22:18:11 +03:00
• 16d6aa15f1 record unittest name in process replay (#9731) qazalandGitHub 2025-04-05 01:39:48 +08:00
• 354db961c6 viz refactor to prep for profiler [pr] (#9739) qazalandGitHub 2025-04-04 17:13:14 +08:00
• fe998798fb linearizer failure test for OUT OF BOUNDS ACCESS (#9738) chenyuandGitHub 2025-04-04 03:48:43 -04:00
• 8b5a523743 fix minimum length in pattern matcher (#9736) George HotzandGitHub 2025-04-04 14:57:01 +08:00
• 640ff681c3 rename bert script to 8xMI300X (#9734) chenyuandGitHub 2025-04-03 23:36:24 -04:00
• b719aa1fb0 only check once for divisible fold lengths (#9732) George HotzandGitHub 2025-04-04 11:27:34 +08:00
• 926b0bcc57 cache folded upcast [pr] (#9733) George HotzandGitHub 2025-04-04 11:23:19 +08:00
• 8206c7281e move const multiply after REDUCE (#9730) George HotzandGitHub 2025-04-04 11:07:46 +08:00
• 6b3480ec70 update mi300x bert haparams (#9716) chenyuandGitHub 2025-04-03 22:30:00 -04:00
• cac8bcf8b5 use Ops.REDUCE (#9721) George HotzandGitHub 2025-04-04 10:14:34 +08:00
• 949459fdd6 jit: fix deallocate on unallocated buffers in free_intermediates (#9699) nimlgenandGitHub 2025-04-03 22:32:51 +07:00
• 52a8ecb15e record unittest location in process replay [pr] (#9727) qazalandGitHub 2025-04-03 20:50:09 +08:00
• ac713e04db ONNX add output shape validation (#9720) geohotstanandGitHub 2025-04-03 17:44:53 +08:00
• 7dadbf3697 insert float() in bert acc (#9726) chenyuandGitHub 2025-04-03 05:44:09 -04:00
• 79145e3d40 cleanup truncate_bf16 [pr] (#9725) chenyuandGitHub 2025-04-03 05:43:49 -04:00
• bc2d86195e increase test tolerance (#9719) Ignacio SicaandGitHub 2025-04-03 15:24:09 +08:00
• 1d25844d44 Revert "disable CI red llama 3 4 gpu beam (#9690)" (#9709) chenyuandGitHub 2025-04-03 02:34:39 -04:00
• 49dafe6d43 add gc tests [pr] (#9718) George HotzandGitHub 2025-04-03 14:08:32 +08:00
• bc91fffc5d fix gated store with index in python backend (#9703) Ignacio SicaandGitHub 2025-04-03 12:48:28 +08:00
• f2bd65ccfc delete Ops.EMPTY and Tensor._metaop (#9715) qazalandGitHub 2025-04-03 12:29:02 +08:00
• 5c7b549eab use functools.cache instead of lru_cache(None) [pr] (#9714) George HotzandGitHub 2025-04-03 11:47:13 +08:00
• bbd13191f4 cleanup tensor BIND + remove outdated comments in tensor.py [pr] (#9712) qazalandGitHub 2025-04-03 11:21:53 +08:00
• e1d7e47cca fix ONNX IsInf unintended dtype promotion (#9711) geohotstanandGitHub 2025-04-03 10:46:15 +08:00
• 11ae254dc5 construct BUFFER UOps directly when device in known [pr] (#9710) qazalandGitHub 2025-04-03 10:41:44 +08:00
• 1714fc3ba4 start work on speed [pr] (#9707) George HotzandGitHub 2025-04-03 10:39:01 +08:00
• 0f1ffc2050 hotfix: cat tests 2048 instead of 256 geohot 2025-04-03 10:37:56 +08:00
• 5bd485c027 Fix double SDMA_OP_FENCE (#9705) uuuvnandGitHub 2025-04-03 06:43:37 +05:00
• a6fec2f5ae dev_run for bert on mi300x (#9706) chenyuandGitHub 2025-04-02 21:12:55 -04:00
• d96b4983ac amd: support rdna4 in runtime again (#9702) nimlgenandGitHub 2025-04-03 01:19:23 +07:00
• 2d6d8b7355 add bf16 mfma support (#9695) Ignacio SicaandGitHub 2025-04-02 21:44:49 +08:00
• a6733f519f dsp: make relro sections contiguous (#9701) nimlgenandGitHub 2025-04-02 18:02:16 +07:00
• ea5caefef0 gep should look at count, not vcount (#9698) George HotzandGitHub 2025-04-02 18:10:57 +08:00
• f72a87fd0e add proper support for Ops.IGNORE to remove store masks (#9692) George HotzandGitHub 2025-04-02 16:38:01 +08:00
• 3b8d923692 remove skip LLVM in test_div_int (#9686) chenyuandGitHub 2025-04-02 04:15:00 -04:00
• bc3bfcbad4 update install gpuocelot (#9693) chenyuandGitHub 2025-04-02 04:10:34 -04:00
• e78e8722dc Revert "LDS noop and spec (#9669)" (#9691) 3 people 2025-04-02 15:31:32 +08:00
• 4514fd91c1 more stuff from DSP (#9689) George HotzandGitHub 2025-04-02 15:27:48 +08:00
• 6a5eacba8b disable CI red llama 3 4 gpu beam (#9690) chenyuandGitHub 2025-04-02 03:19:09 -04:00
• 876a8be97a Debug env var breakdown (#9663) Ignacio SicaandGitHub 2025-04-02 14:34:07 +08:00
• 6f812d3f2f fixes from the dsp branch + 12500 lines (#9683) George HotzandGitHub 2025-04-02 13:07:17 +08:00
• c20f112e9f example test use z3 to verify valid simplification (#9684) chenyuandGitHub 2025-04-02 01:05:52 -04:00
• bca0c85193 skip CI CPU test_data_parallel_resnet_train_step (#9685) chenyuandGitHub 2025-04-02 01:04:54 -04:00
• bb94f13e58 add RECORD_TRACEBACKS=1 option to process replay (#9679) qazalandGitHub 2025-04-02 11:58:27 +08:00
• 3acc1b928a minor div_and_mod_folding cleanup [pr] (#9681) chenyuandGitHub 2025-04-01 23:51:36 -04:00
• c672716b38 improve vmin/vmax for IDIV (#9678) chenyuandGitHub 2025-04-01 23:16:01 -04:00
• 8dd88ad476 don't div_and_mod_folding for negative numerator with remainder (#9674) chenyuandGitHub 2025-04-01 16:26:23 -04:00
• 0e34f9082e helper functions for cstyle div mod [pr] (#9673) chenyuandGitHub 2025-04-01 08:06:56 -04:00
• eee0dcc37a merge viz back into one file (#9672) qazalandGitHub 2025-04-01 19:52:02 +08:00
• 870b545ace LDS noop and spec (#9669) Ignacio SicaandGitHub 2025-04-01 18:44:55 +08:00
• 609a006242 AMDComputeQueue.wreg (#9628) uuuvnandGitHub 2025-04-01 15:01:33 +05:00
• fa373e15a3 hotfix: NULL=1 Buffer does not have _buf (#9661) qazalandGitHub 2025-04-01 17:43:55 +08:00
• 3e2f42c2e8 autogen: remove am headers from extra (#9666) nimlgenandGitHub 2025-04-01 14:45:30 +07:00
• cfad139189 bump assembly debug to 7 (#9662) Ignacio SicaandGitHub 2025-04-01 11:51:33 +08:00
• ac533e89a2 remove duplicated ast print (#9660) Ignacio SicaandGitHub 2025-04-01 10:29:24 +08:00
• 846ef84cda move uops print to debug >= 6 (#9659) Ignacio SicaandGitHub 2025-04-01 10:29:09 +08:00
• d9af4cfc1b AMD_LLVM: tensor cores support (#9613) 3 people 2025-04-01 09:56:27 +08:00
• 1658eb4e63 always fit fresh viz graph into view [pr] (#9657) qazalandGitHub 2025-04-01 09:34:26 +08:00
• a1ee4d587f Fix test_ops for tiny backend (#9302) 3 people 2025-04-01 06:43:09 +05:30
• a0b4465412 bring GroupOp.Meta back (#9656) qazalandGitHub 2025-04-01 01:02:29 +08:00
• f277f407f2 remove smem_prefix_for_cast for amd (#9651) Ignacio SicaandGitHub 2025-03-31 23:03:35 +08:00
• f7cb2e8da3 bert dev_beam for mi300x box (#9648) chenyuandGitHub 2025-03-31 08:35:51 -04:00
• 5171b098e5 merge_double_reduce without asserts [pr] (#9650) qazalandGitHub 2025-03-31 19:17:05 +08:00
• 1444069c09 Uppercase K for dimension and lowercase k for kernel in linearizer tc helper test (#9649) Ignacio SicaandGitHub 2025-03-31 19:05:36 +08:00
• baa67fd124 Uppercase N and M (standalone syntax change) (#9647) Ignacio SicaandGitHub 2025-03-31 18:45:30 +08:00
• aca0f1befb print idx when OUT OF BOUNDS ACCESS (#9646) chenyuandGitHub 2025-03-31 06:12:44 -04:00
• e2d9322d21 torch backend: partial fix for strided related test fails (#9642) Priyank PatelandGitHub 2025-03-31 02:45:18 -07:00
• 76c1b1edf6 viz kernel list cleanup (#9643) qazalandGitHub 2025-03-31 15:53:39 +08:00
• e4c545b396 linearizer fix from dsp branch (#9641) George HotzandGitHub 2025-03-31 14:26:39 +08:00
• ec405b919f Revert "Revert "do not block gc in UOp.toposort (#9623)" (#9624)" (#9639) George HotzandGitHub 2025-03-31 14:03:38 +08:00
• 49b1c46d16 good changes from the dsp branch (#9638) George HotzandGitHub 2025-03-31 13:02:53 +08:00
• 9d67d3a2f3 simpler viz codeblocks (#9636) qazalandGitHub 2025-03-31 11:48:35 +08:00
• 60eb0c4ed7 exclude slow tests on PYTHON (#9634) chenyuandGitHub 2025-03-30 22:55:05 -04:00
• 5012ba3f04 cumalu touchup [pr] (#9632) chenyuandGitHub 2025-03-30 22:43:11 -04:00
• d8d7ac1bb1 fix bert free_intermediates (#9633) chenyuandGitHub 2025-03-30 22:42:52 -04:00
• ff984c807d hotfix: less lines for viz helpers (#9631) qazalandGitHub 2025-03-31 10:10:34 +08:00
• c206a7ae6d refactor viz state updates (#9630) qazalandGitHub 2025-03-31 09:54:54 +08:00
• 6652003839 Add cumprod to Tensor (#9629) 3 people 2025-03-31 03:49:18 +02:00
• d52e91db7b ONNX ops clean ups (#9622) 3 people 2025-03-31 09:39:22 +08:00
• 962c0f65f8 Fix generate_am (#9626) uuuvnandGitHub 2025-03-30 22:15:44 +05:00
• 2a4247b8c2 RDNA 3.5 support (#9627) uuuvnandGitHub 2025-03-30 22:15:20 +05:00
• a08b07b4da Bump onnx==1.17.0 (#9618) 3 people 2025-03-30 15:21:51 +08:00
• 7ef02d0e1c Revert "do not block gc in UOp.toposort (#9623)" (#9624) qazalandGitHub 2025-03-30 12:57:06 +08:00
• f1a35bbb54 do not block gc in UOp.toposort (#9623) qazalandGitHub 2025-03-30 11:54:17 +08:00
• 54e1e59b44 am: rdna 4 support (#9621) nimlgenandGitHub 2025-03-29 23:16:27 +07:00
• 118bd1cbed hotfix: amd imports (#9620) nimlgenandGitHub 2025-03-29 20:19:53 +07:00
• 5908b89f71 MI300X support (WIP) (#9585) uuuvnandGitHub 2025-03-29 16:46:42 +05:00
• 77f0d09ecf hotfix: HIP supports parallel BEAM search geohot 2025-03-29 11:49:53 +08:00
• 162f286a0e add a few Tensor method to doc (#9614) chenyuandGitHub 2025-03-28 13:47:16 -04:00
• dd9aae02c3 Refactor ops_amd.py (MI300X prereq) (#9428) uuuvnandGitHub 2025-03-28 22:17:20 +05:00
• 3e1168ff5e am: module import in common (#9615) uuuvnandGitHub 2025-03-28 19:29:34 +05:00
• a8ff85369e cpugraph for dsp (#9601) nimlgenandGitHub 2025-03-28 19:06:31 +07:00
• fa0ebbd237 jit: optimize before pickle (#9611) nimlgenandGitHub 2025-03-28 19:06:09 +07:00
• 392a311312 Revert "add copy button in VIZ code-block (#9605)" (#9610) George HotzandGitHub 2025-03-28 17:05:44 +08:00
• d1e8598c81 add copy button in VIZ code-block (#9605) Harsh NatuskarandGitHub 2025-03-28 14:22:21 +05:30
• b4ea45b4a6 fix viz recenter + worker cleanup (#9607) qazalandGitHub 2025-03-28 15:24:53 +08:00
• 50dee4a7b3 add test for checking const gradients (#9598) Andrew FureyandGitHub 2025-03-27 19:17:37 +00:00