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

• d803a9c7c8 global metadata try 2 (#7367) qazalandGitHub 2024-10-29 14:21:00 +02:00
• 2cfc7b6695 Index everywhere 2 (#7363) George HotzandGitHub 2024-10-29 18:29:40 +07:00
• 7149eabb34 assert set equality in TestTensorMetadata [pr] (#7364) qazalandGitHub 2024-10-29 13:29:29 +02:00
• 0ebdb136e8 revert metadata with graph_rewrite (#7353) (#7362) qazalandGitHub 2024-10-29 13:16:31 +02:00
• f2044cfb22 hotfix: if getenv("RUN_PROCESS_REPLAY") (#7361) qazalandGitHub 2024-10-29 12:51:29 +02:00
• 0af1212164 use assertEqual with new style uops [pr] (#7360) George HotzandGitHub 2024-10-29 17:43:21 +07:00
• 0beb2d8f84 ptx indexing (#7359) George HotzandGitHub 2024-10-29 17:29:44 +07:00
• 572499c71a add indexing to ops_python (#7358) George HotzandGitHub 2024-10-29 17:11:03 +07:00
• 540e4179e7 global UOp to Metadata mapping + inverse DEBUG=2 metadata order [pr] (#7353) qazalandGitHub 2024-10-29 11:12:00 +02:00
• 2fdfcffe4c improve ci speed [pr] (#7357) George HotzandGitHub 2024-10-29 16:00:35 +07:00
• 8fab7b21df everything is load (#7355) qazalandGitHub 2024-10-29 10:47:33 +02:00
• b647fa7514 rename MathTraits to maximum [pr] (#7356) George HotzandGitHub 2024-10-29 15:43:04 +07:00
• 2bf55d8eda make ops more like tensor [pr] (#7352) George HotzandGitHub 2024-10-29 15:23:41 +07:00
• 3989bd2682 idiv + reciprocal [pr] (#7354) George HotzandGitHub 2024-10-29 14:54:19 +07:00
• 3419ae282d VIZ UI improvement: generic function for vscode opener (#7338) 3 people 2024-10-29 00:02:04 -07:00
• c03e1693fc shorter gate folding [pr] (#7350) qazalandGitHub 2024-10-29 08:49:32 +02:00
• 3e8225299c ext gate indexing (#7349) George HotzandGitHub 2024-10-29 13:46:10 +07:00
• 13ea4979d5 VIZ UI improvement: autoscroll kernel list when using arrow buttons (#7344) Bhavya GadaandGitHub 2024-10-28 23:40:42 -07:00
• d9d4dd6756 faster ci [pr] (#7348) George HotzandGitHub 2024-10-29 13:01:44 +07:00
• a5e0f59e41 move autogen to different CI runner [pr] (#7346) George HotzandGitHub 2024-10-29 12:35:22 +07:00
• 4cb236a495 index in cstyle (#7328) George HotzandGitHub 2024-10-29 12:06:26 +07:00
• f55c3dcff8 hotfix: bump ocelot geohot 2024-10-29 12:46:24 +08:00
• 4fe1945df6 llvm if load (#7345) George HotzandGitHub 2024-10-29 10:33:22 +07:00
• 8625dd4eea minor changes reading ops.py (#7343) chenyuandGitHub 2024-10-28 19:04:12 -04:00
• 6021bf87f4 unify T = TypeVar("T") (#7342) chenyuandGitHub 2024-10-28 18:43:44 -04:00
• 293adc141a clean up get_shape [pr] (#7341) chenyuandGitHub 2024-10-28 18:25:37 -04:00
• c398f2467c test uop mul min/max do not have nan in 0*inf (#7340) chenyuandGitHub 2024-10-28 17:52:01 -04:00
• 0843734927 clean up nan handling in transcendental (#7332) chenyuandGitHub 2024-10-28 16:21:49 -04:00
• 75dcd98e79 Fix calculation of vmin and vmax in multiplication when one src is negative and the other src has negative min and positive max (#7333) 3 people 2024-10-28 21:01:46 +01:00
• 603fcc96f2 limit UOps.ALU min/max to non-float only (#7336) chenyuandGitHub 2024-10-28 15:34:19 -04:00
• 32fa297e6c cleaner nan rendering (#7337) ignaciosicaandGitHub 2024-10-28 15:36:36 -03:00
• 00362a117c scheduler bfs renames [pr] (#7335) qazalandGitHub 2024-10-28 18:24:23 +02:00
• d8820644e0 split preschedule from ast rewrite [pr] (#7334) qazalandGitHub 2024-10-28 17:45:09 +02:00
• 6b0e8cb04f remove float_to_bits in transcendental [pr] (#7331) chenyuandGitHub 2024-10-28 10:20:19 -04:00
• b9b28e6883 viz stuff [pr] (#7330) qazalandGitHub 2024-10-28 15:46:18 +02:00
• 9b7e76e508 VIZ UI improvement: resizable and collapsible sidebars (#7317) Bhavya GadaandGitHub 2024-10-28 06:19:43 -07:00
• e46edc22aa use unittest helpers in TestTensorMetadata [pr] (#7329) qazalandGitHub 2024-10-28 12:38:30 +02:00
• 96fcc47e27 touchup abstraction docs (#7327) chenyuandGitHub 2024-10-27 22:29:55 -04:00
• cb5702f170 tiny cleanup to transcendental xexp2 (#7326) chenyuandGitHub 2024-10-27 21:54:20 -04:00
• 4c855ae692 unit test transcendental helpers (#7325) chenyuandGitHub 2024-10-27 19:55:00 -04:00
• 8d9459f281 always run process replay with contextvars (#7323) qazalandGitHub 2024-10-27 20:44:42 +02:00
• adcdaa17bb map BUFFER to Metadata [pr] (#7324) qazalandGitHub 2024-10-27 20:31:04 +02:00
• d634261c51 late buffer uops [pr] (#7322) qazalandGitHub 2024-10-27 19:34:01 +02:00
• cdbe08b94b use UOp.render in colored_shape (#7321) chenyuandGitHub 2024-10-27 11:42:31 -04:00
• 4a03e00aa1 fix llama3 download_model assert (#7320) chenyuandGitHub 2024-10-27 11:20:24 -04:00
• d4d201d87b fixing branch condition on UOps.IF in the ptx renderer (#7315) 5 people 2024-10-27 08:27:38 -04:00
• a410b46c1d unskip test_gated_store_with_if [pr] (#7319) qazalandGitHub 2024-10-27 14:03:12 +02:00
• 3c992250d5 Failing test: different behavior on different devices (#7193) Maximilian WolfandGitHub 2024-10-27 02:53:58 +01:00
• e920f1d663 Llama 3.2 1B load from GGUF (#7295) eliotgoldingandGitHub 2024-10-27 01:29:02 +00:00
• d66fe7a66f fix simplify_valid (#7313) chenyuandGitHub 2024-10-26 14:21:12 -04:00
• 0a4d01f6d4 disable simplify_valid (#7312) chenyuandGitHub 2024-10-26 12:42:48 -04:00
• 293714610a capture beam log runtime errors (#7311) nimlgenandGitHub 2024-10-26 13:59:45 +03:00
• 3c62315aa8 add resnet pf (#7310) nimlgenandGitHub 2024-10-26 13:20:32 +03:00
• 68cd2c0669 nv correct local memory based on device (#7307) nimlgenandGitHub 2024-10-25 22:23:42 +03:00
• 2ddfb9678a update exponent_bias in transcendental (#7304) chenyuandGitHub 2024-10-25 10:45:49 -04:00
• e7cd21c5e3 remove custom render in test_simplify_valid_idx (#7303) chenyuandGitHub 2024-10-25 10:20:26 -04:00
• 4688c01e3e transcendental cleanups (#7301) chenyuandGitHub 2024-10-25 09:30:25 -04:00
• aadf688aeb order flipper as *normal* rewrite rule (#7300) George HotzandGitHub 2024-10-25 20:28:30 +07:00
• 3c31497f55 instant isn't actually used [pr] (#7299) George HotzandGitHub 2024-10-25 20:01:29 +07:00
• 199a991237 line reduction [pr] (#7296) George HotzandGitHub 2024-10-25 17:05:09 +07:00
• 4fed358511 hotfix: timeouts to 20 minutes. better no stats update than a red x geohot 2024-10-25 16:31:52 +08:00
• dc3148c677 hotfix: minor speed increase + stable diffusion relax geohot 2024-10-25 16:27:21 +08:00
• 4812801aa6 try for canonical order (#7286) George HotzandGitHub 2024-10-25 15:04:54 +07:00
• d3500af71b move consts last in uop toposort (#7290) George HotzandGitHub 2024-10-25 13:58:48 +07:00
• e3c9c94896 Revert "move anything that isn't bfs [pr] (#7273)" (#7289) qazalandGitHub 2024-10-25 09:38:30 +03:00
• 0b47eca085 schedule.py reorders [pr] (#7285) qazalandGitHub 2024-10-25 09:30:23 +03:00
• 004af512e6 try all matches in the function (#7288) George HotzandGitHub 2024-10-25 13:17:04 +07:00
• bcf0537653 canonicalize the order prereqs (#7283) George HotzandGitHub 2024-10-25 10:37:51 +07:00
• 603d637105 split to fuse.py and schedule.py [pr] (#7284) qazalandGitHub 2024-10-25 06:17:24 +03:00
• 698457c5ce big graph ScheduleContext [pr] (#7282) qazalandGitHub 2024-10-25 05:58:23 +03:00
• b805711f86 move anything that isn't bfs [pr] (#7273) qazalandGitHub 2024-10-25 05:34:21 +03:00
• 6dc7d3c949 instant uop rules [pr] (#7263) George HotzandGitHub 2024-10-25 09:32:45 +07:00
• 90f720d703 limit idiv by neg bound to only if s0 is non-negative [pr] (#7277) chenyuandGitHub 2024-10-24 15:46:50 -04:00
• d4c94d0d32 disable llama 1 4gpu and 6gpu benchmark (#7276) chenyuandGitHub 2024-10-24 14:19:22 -04:00
• e6929f2402 RUN_PROCESS_REPLAY=0 on llama 70B and resnet training (#7272) chenyuandGitHub 2024-10-24 12:09:54 -04:00
• b777cfdcba update test_max_simplify_and_cancel (#7270) chenyuandGitHub 2024-10-24 10:29:05 -04:00
• d3953c6c55 split the preload path [pr] (#7271) qazalandGitHub 2024-10-24 17:28:03 +03:00
• acad11ea8e minor cleanup to View add [pr] (#7247) chenyuandGitHub 2024-10-24 09:18:47 -04:00
• 98f8d0ccf9 nv limit max local memory with envvar (#7265) nimlgenandGitHub 2024-10-24 16:01:50 +03:00
• f20b651ee0 prescheduling refactor from big graph [pr] (#7268) qazalandGitHub 2024-10-24 14:55:07 +03:00
• 2e6ec43c49 hotfix: become the latest for process replay geohot 2024-10-24 19:02:59 +08:00
• 9a3d498d9c with commutative hack, uops can change. fix that (#7266) George HotzandGitHub 2024-10-24 17:50:23 +07:00
• d482d927a8 hotfix: nobody uses [run_process_replay] [pr] (#7264) qazalandGitHub 2024-10-24 13:37:29 +03:00
• fa5dc7857a assign toposort with big graph, bfs [pr] (#7242) qazalandGitHub 2024-10-24 13:09:01 +03:00
• 4d081eb560 double mod is single mod (#7262) George HotzandGitHub 2024-10-24 17:02:51 +07:00
• e4631a47f4 symbolic arange support (#7252) George HotzandGitHub 2024-10-24 16:55:53 +07:00
• 23b26d40d5 small max rules + windows VIZ (#7261) George HotzandGitHub 2024-10-24 16:43:39 +07:00
• 415186da3c Revert "some rules to simplify max (#7258)" (#7260) George HotzandGitHub 2024-10-24 16:15:52 +07:00
• 93934c2160 early assert cyclic read [pr] (#7259) qazalandGitHub 2024-10-24 11:51:12 +03:00
• b56fab54ea some rules to simplify max (#7258) George HotzandGitHub 2024-10-24 15:27:21 +07:00
• a7be9dfd71 leftover lru_cache on UPat [pr] (#7257) George HotzandGitHub 2024-10-24 15:11:24 +07:00
• 532b7b018c add smin/smax (#7253) George HotzandGitHub 2024-10-24 15:10:49 +07:00
• de7b9d7c42 improve pre-commit [pr] (#7256) George HotzandGitHub 2024-10-24 14:38:47 +07:00
• b1a30677fe add some tiny tests that should pass everywhere [pr] (#7254) George HotzandGitHub 2024-10-24 13:38:46 +07:00
• 63048ad880 don't recreate COMMUTATIVE the other way (#7255) George HotzandGitHub 2024-10-24 13:38:29 +07:00
• 1315b8909a strict mode triggers on beam timeout [pr] (#7250) George HotzandGitHub 2024-10-24 10:37:57 +07:00
• 9f32a6f496 Revert "move metal tc check to renderer [pr] (#7248)" (#7251) George HotzandGitHub 2024-10-24 09:57:09 +07:00
• 72ddcdb4d1 move metal tc check to renderer [pr] (#7248) George HotzandGitHub 2024-10-24 09:38:57 +07:00
• 451c043552 narrow return type of bool, int, float on UOp [pr] (#7246) chenyuandGitHub 2024-10-23 21:06:43 -04:00
• 9f370cccb3 minor cleanups in apply_opt [pr] (#7243) chenyuandGitHub 2024-10-23 18:21:00 -04:00