Commit Graph

  • e555748807 test rangeify const folding (#12200) chenyu 2025-09-15 20:03:48 -04:00
  • f732f66709 rangeify test_nn almost pass (#12198) chenyu 2025-09-15 17:49:20 -04:00
  • 82e037aad5 ci test.yml updates (#12197) chenyu 2025-09-15 17:09:02 -04:00
  • 146c31586d split RANGEIFY ci (#12196) chenyu 2025-09-15 15:41:10 -04:00
  • df1c183e46 Revert "more llvm intrinsics (#11961)" (#12194) chenyu 2025-09-15 13:56:43 -04:00
  • d01e3d7719 more llvm intrinsics (#11961) b1tg 2025-09-16 01:05:23 +08:00
  • b63bd02969 update runtime docs (#12191) nimlgen 2025-09-15 17:46:20 +03:00
  • 57e8bf61e8 viz: fix Specificity for rect styling (#12190) qazal 2025-09-15 17:33:37 +03:00
  • 72e010d816 fix rangeify ci (#12189) chenyu 2025-09-15 10:24:57 -04:00
  • f1bd06134d test fuse with RANGEIFY=2 (#12187) qazal 2025-09-15 15:51:23 +03:00
  • ef0ef705fe viz: remove async from event listener (#12186) qazal 2025-09-15 15:08:28 +03:00
  • d8855ec266 viz/serve.py cleanups (#12185) qazal 2025-09-15 13:43:26 +03:00
  • b8a74c1569 cpu: add disassembler err message (#12184) qazal 2025-09-15 13:29:44 +03:00
  • a388d2cb1a remove PROFILE=1 option, it's just VIZ=1 [pr] (#12176) qazal 2025-09-15 12:51:50 +03:00
  • 65397bfdeb set testpath on pytest (#12183) George Hotz 2025-09-15 16:13:05 +08:00
  • ccd753e1aa set testpath on pytest set_testpath George Hotz 2025-09-15 14:29:30 +08:00
  • ae0edc8a67 renumber ranges (#12182) George Hotz 2025-09-15 13:03:39 +08:00
  • e1fef895b1 don't hardcode weights path (#12171) hooved 2025-09-15 00:33:47 -04:00
  • 3a9db08b49 download data and ckpts for sd train/eval (#12170) hooved 2025-09-15 00:31:45 -04:00
  • bdb3afd566 failed test case for symbolic pad (#12179) chenyu 2025-09-15 00:25:21 -04:00
  • 9fcc87761e enable rangeify const folding (#12181) George Hotz 2025-09-15 12:02:19 +08:00
  • 34ef448331 enable rangeify const folding rangeify_const_folding George Hotz 2025-09-15 11:50:13 +08:00
  • 1353250b6c tags on bufferize are the tensor tags (#12180) George Hotz 2025-09-15 11:46:03 +08:00
  • 60d7db093e delete bufferized consts + output noops (#12163) George Hotz 2025-09-15 11:07:44 +08:00
  • 525c20dc7e viz: remove unused runtime_stats feature (#12177) qazal 2025-09-15 02:53:05 +03:00
  • 75ff9b7a9a viz: add buffer lifetime to tooltip (#12175) qazal 2025-09-15 02:33:50 +03:00
  • 15b166ce6d bump test_module_runs to 30 seconds (#12174) chenyu 2025-09-14 16:48:40 -04:00
  • 943236ef74 move cast pat out of symbolic_simple (#11945) ttomsa 2025-09-14 20:39:48 +01:00
  • 25b1bc8eff added top k sampling to examples/mamba (#12061) Steven Shi 2025-09-14 15:27:34 -04:00
  • 34a05b31fe Fix advanced tensor indexing setitem (#12128) Shun Usami 2025-09-14 12:22:40 -07:00
  • d09c0f28c5 increase test_module_runs (#12173) chenyu 2025-09-14 15:19:21 -04:00
  • 12a910f1d2 update torch 2.8 (#12172) chenyu 2025-09-14 15:19:03 -04:00
  • 98ecab7563 remove ml_dtypes (#12169) chenyu 2025-09-14 14:20:05 -04:00
  • 02054b53fe remove tests that pre date the uop spec (#12168) qazal 2025-09-14 18:47:42 +03:00
  • 1591e4f66b update outbufs selection in test_linearizer [pr] (#12166) qazal 2025-09-14 13:46:49 +03:00
  • d1ae30f7ef hcq: do not spam with errors in -m device (#12150) nimlgen 2025-09-14 10:56:59 +03:00
  • d5bc27797b fix some multitensor on rangeify (#12162) George Hotz 2025-09-14 14:31:57 +08:00
  • 4b7904eca9 add cpu support for riscv64 (#12136) Meng Zhuo 2025-09-14 11:40:58 +08:00
  • bcafa72b7f use tags instead of graph_rewrite_map in rangeify (#12110) George Hotz 2025-09-14 11:39:01 +08:00
  • d2316ba91a don't validate output in sdxl with fakeweights (#12160) chenyu 2025-09-13 21:47:51 -04:00
  • b1d1816f43 device: fix envvars (#12159) nimlgen 2025-09-13 23:38:09 +03:00
  • 19d9d29b7e device: compilers in tinygrad.device (#12151) nimlgen 2025-09-13 21:45:29 +03:00
  • 6410dcb7c2 viz: less verbose render loop (#12158) qazal 2025-09-13 19:04:37 +03:00
  • 92df52d79a make method_cache account for compiler (#12156) nimlgen 2025-09-13 17:00:11 +03:00
  • 0c392089d9 update mypy (#12155) chenyu 2025-09-13 09:48:38 -04:00
  • fbca6183ad do not launch BEAM when opts_to_apply exists [pr] (#12152) qazal 2025-09-13 14:57:46 +03:00
  • b2a95d32bb check clSetKernelArg (#12149) George Hotz 2025-09-13 17:24:55 +08:00
  • e0bc99b6c9 check clSetKernelArg check_clsetarg George Hotz 2025-09-13 16:05:29 +08:00
  • 0695e322a8 fix android cpu device (#12148) George Hotz 2025-09-13 15:42:04 +08:00
  • e3a3764917 delete fold_unrolled_divs (#12146) Sieds Lykles 2025-09-13 03:09:36 +02:00
  • 51ed6e94b2 AxisType __repr__ method (#12145) Sieds Lykles 2025-09-13 01:15:38 +02:00
  • 0757a9a819 add pytest-timeout of 3 min per item (#12144) Sieds Lykles 2025-09-13 00:48:41 +02:00
  • 2fc0bd150b Arange overflow raises error and one_hot upcast (#11975) Sieds Lykles 2025-09-13 00:18:25 +02:00
  • aac3dceaf6 merge two PYTHON backend ci job (#12143) chenyu 2025-09-12 17:36:46 -04:00
  • a12d0933c1 fix vec dtype in fast idiv (#12080) ttomsa 2025-09-12 22:00:43 +01:00
  • 25091951ba update test/models (#12142) chenyu 2025-09-12 16:43:28 -04:00
  • 62376c8b2b update store load noop pattern to use Invalid (#12141) Sieds Lykles 2025-09-12 22:25:53 +02:00
  • 647965fb09 test_train cleanup (#12140) chenyu 2025-09-12 13:21:30 -04:00
  • 0fad07c684 viz serve default path (#12139) chenyu 2025-09-12 11:32:44 -04:00
  • 81e33b8439 system: cpu memory mappings are uncached (#12137) nimlgen 2025-09-12 13:28:25 +03:00
  • 68b0ad05a4 viz: format tuple tags (#12135) qazal 2025-09-12 11:36:53 +03:00
  • e80c8a7548 merge TestIndexing with TestSchedule + remove duplicate tests (#12134) qazal 2025-09-12 10:35:14 +03:00
  • 3128e19003 failing assign test failing_assign George Hotz 2025-09-12 12:53:48 +08:00
  • b5a3b8de20 remove where on gated load if gates are the same (#12129) Sieds Lykles 2025-09-12 06:52:35 +02:00
  • a2f502b89e fix rangeify=1 ops on GPU (#12130) George Hotz 2025-09-12 11:17:37 +08:00
  • 0766616962 isolate the const hacks in the old kernelize (#12126) George Hotz 2025-09-12 08:35:35 +08:00
  • 1f3950a484 Invalid idx (#12067) Sieds Lykles 2025-09-12 01:42:02 +02:00
  • 544eb2c402 clean up test_scatter_reduce (#12125) chenyu 2025-09-11 16:36:58 -04:00
  • 9ad6a56d17 smaller test_simple_reduce (#12124) chenyu 2025-09-11 15:45:38 -04:00
  • e5ef9ec5b1 remove IGNORE_OOB=0 in ci tests (#12117) chenyu 2025-09-11 15:05:04 -04:00
  • 3a83b56da5 fix test_dequantization_mxfp4 (#12123) chenyu 2025-09-11 14:22:06 -04:00
  • 520e2e0727 actually run unit tests in ci MacOS (unit) (#12122) chenyu 2025-09-11 13:32:30 -04:00
  • acb700fc26 ci: fix ptx env (#12120) nimlgen 2025-09-11 19:42:15 +03:00
  • 20cd7177de delete test_bert_fuse_arange (#12121) chenyu 2025-09-11 12:35:51 -04:00
  • b07f962058 split metal model tests (#12119) chenyu 2025-09-11 12:20:12 -04:00
  • 66593f135f remove duplicated test_real_world (#12118) chenyu 2025-09-11 11:57:14 -04:00
  • e76211fcbc viz: specify all rect styles in parent (#12115) qazal 2025-09-11 13:48:59 +03:00
  • 400ad93892 ci: gate boost paths for macos only (#12114) nimlgen 2025-09-11 12:48:34 +03:00
  • 3ef0e5e01e rangeify: use Ops.REALIZE and not Ops.CONTIGUOUS if it's added by system (#12111) George Hotz 2025-09-11 11:56:59 +08:00
  • 52ebed991e change dtype promo lattice when fp8s is supported (#12088) b1tg 2025-09-11 10:09:11 +08:00
  • d4eba5800d rangeify cost function infrastructure (#12091) George Hotz 2025-09-11 07:19:53 +08:00
  • 78610b681e viz: light up children (#12107) qazal 2025-09-11 01:28:01 +03:00
  • 3989f5b559 Revert "Simplify valid in symbolic (#12104)" (#12108) Sieds Lykles 2025-09-10 23:36:40 +02:00
  • f52b977cab Revert "Simplify valid in symbolic (#12104)" revert-12104-simplify_valid_in_symbolic Sieds Lykles 2025-09-10 23:35:13 +02:00
  • 73d479a016 Simplify valid in symbolic (#12104) Sieds Lykles 2025-09-10 23:26:19 +02:00
  • e306650d39 remove GPUDevice (#12106) chenyu 2025-09-10 16:35:00 -04:00
  • d8a7a1c9c7 BUFFERIZE shape should be each range, not the product (#12105) George Hotz 2025-09-11 04:02:24 +08:00
  • 3730172c10 cleanup cast_folding (#12101) Sieds Lykles 2025-09-10 21:30:20 +02:00
  • 0e266f376c ops_gpu -> ops_cl (#12103) chenyu 2025-09-10 15:15:48 -04:00
  • 0599e86186 replace hardcoded GPU in llama debug msg (#12102) chenyu 2025-09-10 13:56:40 -04:00
  • 5a84d86db7 viz: fix buffer tooltip offset (#12100) qazal 2025-09-10 20:12:20 +03:00
  • fb96394ff5 auto-select available compilers (#12094) nimlgen 2025-09-10 19:52:01 +03:00
  • bb67829e99 raise KernelOptError in TC _apply_tc_opt (#12099) chenyu 2025-09-10 12:32:19 -04:00
  • 84b249ef0e move simplify reduce out of devectorizer (#12098) George Hotz 2025-09-10 21:24:57 +08:00
  • 5d66a2d885 viz: refactor range clipping (#12097) qazal 2025-09-10 16:23:46 +03:00
  • 9789337722 early reduce simplify (#12046) George Hotz 2025-09-10 21:02:46 +08:00
  • 21e6926a6a HostLLVMCompiler -> CPULLVMCompiler (#12096) nimlgen 2025-09-10 14:04:16 +03:00
  • 551560b87c do not use getenv('PTX') in tests (#12095) nimlgen 2025-09-10 14:04:07 +03:00
  • 0e420e68b4 delete axis_is_masked (#12092) Sieds Lykles 2025-09-10 05:26:19 +02:00
  • ef53a6fc19 one call to hc opt (#12074) George Hotz 2025-09-09 20:18:18 -07:00