Commit Graph

  • b63e5a7568 viz: full range x axis scroll (#13459) qazal 2025-11-26 21:28:07 +08:00
  • c12e218751 viz: double click on INST wave (#13458) qazal 2025-11-26 21:12:40 +08:00
  • e9cb738c7a viz: event sidebar cleanup (#13457) qazal 2025-11-26 19:47:15 +08:00
  • 2a3b665972 viz: initial zoom at first event (#13456) qazal 2025-11-26 16:42:06 +08:00
  • b2af92c821 fix HCQGraph.__del__ bug when finalizing (#13298) Christopher Milan 2025-11-25 23:33:48 -05:00
  • 8c1e2a42fd viz: start work on profiler speed (#13455) qazal 2025-11-26 07:54:04 +08:00
  • ffc31a23f4 tk mi350 (#13288) wozeparrot 2025-11-25 15:49:44 -08:00
  • 436ab6bfc7 nv: use opt mutliple vaspaces (#13453) nimlgen 2025-11-25 23:10:21 +03:00
  • 7238df7a94 viz: cleanup sort_fn (#13454) qazal 2025-11-26 04:10:10 +08:00
  • 5520f1fb0b viz: per cu timeline (#13451) qazal 2025-11-26 00:05:20 +08:00
  • 4a9562e353 viz: draw markers on top (#13449) qazal 2025-11-25 17:27:01 +08:00
  • 5373fd2d66 add user device (#13447) George Hotz 2025-11-24 23:25:45 -08:00
  • 241e533451 toposort recursive_property is faster (#13446) George Hotz 2025-11-24 22:29:15 -08:00
  • 9d8a2bd212 toposort recursive_property is faster topoprop George Hotz 2025-11-24 22:09:32 -08:00
  • ee3ed9e646 wip junk George Hotz 2025-11-24 20:03:12 -08:00
  • 62f98ef817 fused optim George Hotz 2025-11-24 19:51:07 -08:00
  • f6dcb9a777 why did fakedata have fakeweights? George Hotz 2025-11-24 19:29:42 -08:00
  • dfaaeb0720 improve llama trainer George Hotz 2025-11-24 19:11:14 -08:00
  • 8e8fec408e fix n^2 _apply_map_to_tensors [pr] (#13443) George Hotz 2025-11-24 18:59:16 -08:00
  • 249553a119 tinyfs tweaks (#13444) wozeparrot 2025-11-24 18:07:32 -08:00
  • f46bc31156 tk: start and step in range (#13442) wozeparrot 2025-11-24 15:43:24 -08:00
  • cc5e6323ac stable diffusion profiling (#13441) George Hotz 2025-11-24 15:25:45 -08:00
  • 18cfb54736 amd: a bit better se limiting (#13440) nimlgen 2025-11-24 21:51:47 +03:00
  • 2d53029be3 Whisper less flaky tests (#13435) C T 2025-11-24 19:50:49 +02:00
  • 2a9bd12700 sqtt: add occupancy events to the timeline (#13430) qazal 2025-11-24 22:28:05 +08:00
  • 63a931ff76 Symbolic divisor fuzzer (#13433) Sieds Lykles 2025-11-23 20:29:32 +01:00
  • 677db34eba nv: cleanup map flags (#13434) nimlgen 2025-11-23 19:54:52 +03:00
  • 712c7a6448 sqtt loader cleanups from the occupancy branch (#13431) qazal 2025-11-23 21:50:34 +08:00
  • 9d7a17ee39 beautiful SQTT_PARSE=1 with color (#13428) George Hotz 2025-11-23 01:05:14 -08:00
  • 474a631877 viz: align left offset for nested items (#13420) qazal 2025-11-23 14:22:51 +08:00
  • da0aa57a3b add cu parsing to attempt_sqtt_parse George Hotz 2025-11-22 22:08:18 -08:00
  • 320ed78803 can view wave timeline with SQTT_ITRACE_SE_MASK=0 (#13427) qazal 2025-11-23 13:55:47 +08:00
  • c1838c71fc display service name typo (#13426) Pranil 2025-11-22 20:49:56 -08:00
  • 5110409339 continue work on parse sqtt, enable with SQTT_PARSE (#13425) George Hotz 2025-11-22 19:03:17 -08:00
  • 92170d0ff1 lil op cleanup (#13424) George Hotz 2025-11-22 15:21:15 -08:00
  • 423b76a852 improve sqtt format parser (saturday coffee shop project) (#13419) George Hotz 2025-11-22 15:04:10 -08:00
  • 9d6cf3472e remove op/sentinel George Hotz 2025-11-22 15:01:47 -08:00
  • 310da2a201 remove hashFiles in setup-tinygrad (#13423) Christopher Milan 2025-11-22 17:47:10 -05:00
  • c14033e10f viz: faster startup time with SQTT=1 (#13337) qazal 2025-11-22 22:02:30 +08:00
  • 1655fdb6de viz: cleanup sqtt loader (#13417) qazal 2025-11-22 20:10:23 +08:00
  • 903eec3754 fix sz.py tinygrad import in ci (#13418) qazal 2025-11-22 19:20:26 +08:00
  • 3a42680e22 amd: pmc generic arch for gfx10+ (#13407) nimlgen 2025-11-22 12:31:23 +03:00
  • 1f8b24a6b9 track flag count and op count (#13416) George Hotz 2025-11-21 22:46:33 -08:00
  • 4c0f4226b9 delete the PRECAST op [p] (#13415) George Hotz 2025-11-21 21:47:14 -08:00
  • 1f648bb1ba feat: reenable mobilenetv2 dsp (#13320) wozeparrot 2025-11-21 15:21:49 -08:00
  • 054477a44f remove full_symbolic in simplify (#13413) chenyu 2025-11-21 15:04:00 -05:00
  • cb29265f23 add test that shows the validhack regression with bad rewrite order (#13411) chenyu 2025-11-21 13:48:30 -05:00
  • fdfe83880b viz: unique sqtt wave names (#13410) qazal 2025-11-22 02:43:31 +08:00
  • a6c9b4ff6a fix symbolic comments [pr] (#13408) chenyu 2025-11-21 09:18:50 -05:00
  • 114bb94c55 Fix load collapse MAX to ADD (#13406) Sieds Lykles 2025-11-21 12:26:14 +01:00
  • 87c248eafa small cleanups from viz memory usage fixes (#13405) qazal 2025-11-21 17:05:08 +08:00
  • 0de1b24154 viz: SE : CU : SIMD : WAVE in sqtt timeline (#13404) qazal 2025-11-21 15:42:29 +08:00
  • dabb02767f set AMD profile mode with sudo on SQTT or PMC (#13403) George Hotz 2025-11-20 23:19:11 -08:00
  • e1051d00d7 multi like on full_like as well as rand_like (#13402) George Hotz 2025-11-20 20:46:48 -08:00
  • fa3def2f12 call less simplify in simplify_valid_load [pr] (#13401) chenyu 2025-11-20 19:54:22 -05:00
  • 895ec7417e viz: enable mapping function names to colors (#13400) qazal 2025-11-21 06:43:02 +08:00
  • a74f6020d5 track apply map to tensors (#13399) George Hotz 2025-11-20 14:24:55 -08:00
  • 647fde64e6 no sym in pm_reduce [pr] (#13398) chenyu 2025-11-20 16:49:09 -05:00
  • 1313250e0d viz: use system helper for llvm-mca (#13395) qazal 2025-11-21 04:47:25 +08:00
  • de3593957f Revert "Revert "autogen: fix formatting on zero-argument function-like macros…" (#13388) Christopher Milan 2025-11-20 15:36:13 -05:00
  • 1220072328 viz: refactor to generic steps api (#13393) qazal 2025-11-21 04:33:23 +08:00
  • 26ccbf7040 debufferize with symbolic in one pm (#13392) George Hotz 2025-11-20 11:47:03 -08:00
  • c46f608703 top down remove_bufferize (#13391) George Hotz 2025-11-20 11:32:00 -08:00
  • 4043489803 set curl -f in setup-tinygrad (#13389) Christopher Milan 2025-11-20 13:45:47 -05:00
  • 0251a8e628 parse_valid minor cleanup [pr] (#13385) chenyu 2025-11-20 13:15:06 -05:00
  • 0901a40685 Revert "autogen: fix formatting on zero-argument function-like macros (#13386)" (#13387) Christopher Milan 2025-11-20 12:45:35 -05:00
  • 91e289cb14 amd fp8 llvm (#13186) b1tg 2025-11-21 01:35:57 +08:00
  • 1058748440 torch backend: no aten.detach for torch 2.10 compat (#13381) Roelof van Dijk 2025-11-20 18:12:15 +01:00
  • 58d85d4bab autogen: fix formatting on zero-argument function-like macros (#13386) Christopher Milan 2025-11-20 12:11:04 -05:00
  • 9dbc550692 roc: map disassembly to prog name (#13384) qazal 2025-11-20 23:47:19 +08:00
  • ebcdf68bab viz: use content headers for profiler (#13383) qazal 2025-11-20 23:33:16 +08:00
  • 0b0ea4981c hcq: unwrap signals (#13382) nimlgen 2025-11-20 18:12:41 +03:00
  • 9dcd52287a add external_benchmark_pyrender (#13378) qazal 2025-11-20 17:38:28 +08:00
  • cb38c704c3 delete nonfunctional ramp.py George Hotz 2025-11-19 20:43:44 -08:00
  • 8919c994b7 Revert "AxisType.PLACEHOLDER in reshape to do less graph_rewrite (#13373)" (#13375) George Hotz 2025-11-19 19:34:30 -08:00
  • ac7559e33d AxisType.PLACEHOLDER in reshape to do less graph_rewrite (#13373) George Hotz 2025-11-19 19:19:58 -08:00
  • 050682ab40 use invalid_gate consistently [pr] (#13374) chenyu 2025-11-19 22:15:12 -05:00
  • 2c8ad1b419 _apply_movement_op cache cache_reshape George Hotz 2025-11-19 16:10:00 -08:00
  • 821f3771df AxisType.PLACEHOLDER in reshape to do less graph_rewrite George Hotz 2025-11-19 16:04:23 -08:00
  • cb5d827ed9 use buf_target in expand_index bt_ei George Hotz 2025-11-19 15:37:19 -08:00
  • 0dc2ff431d fix: revive torch backend (#13280) Roelof van Dijk 2025-11-20 00:26:50 +01:00
  • 56b2540349 tk: keep extra tile data by replacing uop (#13370) wozeparrot 2025-11-19 15:11:43 -08:00
  • ab7df42c78 bring back fold_divmod_general with bugfix and test [pr] (#13369) George Hotz 2025-11-19 14:51:51 -08:00
  • 986d113024 symbolic fuzz failure (#13367) George Hotz 2025-11-19 14:21:08 -08:00
  • 05ccc69248 Revert "merge to fold_divmod_general [p] (#13359)" George Hotz 2025-11-19 14:18:09 -08:00
  • 90e5752199 Revert "actually merge to fold_divmod_general [pr] (#13363)" George Hotz 2025-11-19 14:18:08 -08:00
  • 8e17bd6791 Revert "add cache to fold_divmod_general (#13365)" George Hotz 2025-11-19 14:18:08 -08:00
  • b5309a5043 add cache to fold_divmod_general (#13365) George Hotz 2025-11-19 13:49:18 -08:00
  • 3d82b83cec actually merge to fold_divmod_general [pr] (#13363) George Hotz 2025-11-19 13:17:56 -08:00
  • a91f00925b remove VECTORIZE and WMMA rules from sym [pr] (#13362) chenyu 2025-11-19 14:51:21 -05:00
  • 7711bbac7f merge to fold_divmod_general [p] (#13359) George Hotz 2025-11-19 11:37:45 -08:00
  • 6fdbd03104 more divmod cleanup [p] (#13358) George Hotz 2025-11-19 10:35:15 -08:00
  • bd88a72149 div and mod to its own file, try 2 [p] (#13357) George Hotz 2025-11-19 10:10:06 -08:00
  • 957cf717e7 Python speed (#13355) George Hotz 2025-11-19 09:03:00 -08:00
  • 82aa943cd4 fix that test python_speed George Hotz 2025-11-19 08:48:49 -08:00
  • fc19ea76b5 clean up threefry rules (#13354) chenyu 2025-11-19 11:48:07 -05:00
  • e16782cf9e Merge branch 'master' into python_speed George Hotz 2025-11-19 08:41:40 -08:00
  • 1c47ee729e fix names of rewrite rules George Hotz 2025-11-19 08:41:34 -08:00
  • a8f9e69bd9 work on python speed George Hotz 2025-11-19 08:34:15 -08:00
  • 385618d45b skip process replay by default (#13353) George Hotz 2025-11-19 08:25:34 -08:00