Commit Graph

  • f6a78a29e0 support einsum trace (#14012) chenyu 2026-01-04 19:27:27 -05:00
  • b52ff63896 fixes George Hotz 2026-01-04 15:48:31 -08:00
  • 404eed6172 assembly/amd: improve tests for asm (#14007) George Hotz 2026-01-04 15:14:08 -08:00
  • 7f7f12d5b4 99% match George Hotz 2026-01-04 15:05:05 -08:00
  • b10ae6958e roundtripping George Hotz 2026-01-04 14:31:40 -08:00
  • f550f9204c fa: failing test for bwd jit (#14009) wozeparrot 2026-01-04 16:57:43 -05:00
  • 10e2c47d52 don't make dtype George Hotz 2026-01-04 13:49:47 -08:00
  • 058816dd92 use tinygrad UOps as DSL George Hotz 2026-01-04 13:40:15 -08:00
  • 28846cb6c4 simpler dsl George Hotz 2026-01-04 13:24:01 -08:00
  • 958bfa1c5b Op George Hotz 2026-01-04 13:05:38 -08:00
  • 23cf30820f more correct George Hotz 2026-01-04 12:52:20 -08:00
  • ea51512f90 CMPLE George Hotz 2026-01-04 12:45:21 -08:00
  • e9664fdf28 dtype is uop George Hotz 2026-01-04 12:31:40 -08:00
  • 5d50281896 Merge remote-tracking branch 'origin/master' into asm_ucode George Hotz 2026-01-04 12:22:52 -08:00
  • cfeeab8485 work George Hotz 2026-01-04 12:22:01 -08:00
  • 7abf4591ba use bitsize on dtype (#14011) George Hotz 2026-01-04 12:16:21 -08:00
  • 2be5f8b688 work George Hotz 2026-01-04 11:57:42 -08:00
  • db9140b8b7 work George Hotz 2026-01-04 11:34:07 -08:00
  • 63f663bd4b progress George Hotz 2026-01-04 10:24:40 -08:00
  • cfb8bf5814 faster image load (#13977) chenyu 2026-01-04 13:09:59 -05:00
  • e38d311f3c simpler George Hotz 2026-01-04 10:02:33 -08:00
  • 8e8ad423a7 post parser George Hotz 2026-01-04 09:16:00 -08:00
  • 7ebda28692 assembly/amd: add CDNA support to asm (#13982) George Hotz 2026-01-04 08:53:56 -08:00
  • acad5d7b30 parser George Hotz 2026-01-04 08:43:41 -08:00
  • 9ef8ae3199 qcode George Hotz 2026-01-04 08:33:57 -08:00
  • 1f96afb1cb getting big George Hotz 2026-01-04 08:05:23 -08:00
  • ad041416ca delete unused rewrite rule [pr] (#14006) chenyu 2026-01-04 09:48:52 -05:00
  • bf356ae996 am: mi300 48bit address space (#14004) nimlgen 2026-01-04 15:19:25 +03:00
  • 606786e152 am: do not sleep for each hive node during resets (#14003) nimlgen 2026-01-04 14:02:11 +03:00
  • 59144c6af6 assembly/amd: start replacing pcode with ucode George Hotz 2026-01-03 23:27:41 -08:00
  • 34ea053b26 assembly/amd: clean up pcode, jit pcode instead of static (#14001) George Hotz 2026-01-04 02:06:15 -05:00
  • 280790e438 Reuse toposort in recursive_property (#13993) kamilisjon 2026-01-04 08:04:13 +02:00
  • 9a9564118c [pr] Delete reverse_toposort (#13987) kamilisjon 2026-01-04 08:03:44 +02:00
  • 8328511808 assembly/amd: make the emu.py code shine (#13996) George Hotz 2026-01-03 23:33:09 -05:00
  • bdb421f13e process_replay: passthrough sink arg for Ops.PROGRAM input (#14000) qazal 2026-01-03 23:09:39 -05:00
  • 66caa9fe1d fix: library linking for fedora systems (#13999) Galax 2026-01-04 02:40:56 +01:00
  • 8003db2a28 test case of NOOP store load folding (#13997) chenyu 2026-01-03 14:39:26 -05:00
  • c1b8644a3f test removing expander rules [pr] (#13994) chenyu 2026-01-03 12:38:01 -05:00
  • 35c2870b1f gate image_conv2d pitch hacks on IMAGE==1 (#13995) Christopher Milan 2026-01-03 09:27:31 -08:00
  • a49924a0e9 hcq: _sleep report status (#13992) nimlgen 2026-01-03 14:28:28 +03:00
  • 3b354bc11f hcq: better queue managment (#13991) nimlgen 2026-01-03 13:11:15 +03:00
  • efb2ae87c6 hcq sync aql (#13756) nimlgen 2026-01-03 12:59:24 +03:00
  • bd55507ee4 RDNA3 fp16 assembly gemm 85 TFLOPS (#13990) qazal 2026-01-03 18:34:23 +09:00
  • 6242a9d151 tk: no global copy and clear ranges (#13988) wozeparrot 2026-01-03 02:45:15 -05:00
  • 9f082e8e25 fa: split kv bwd into 2 kernels (#13981) wozeparrot 2026-01-02 21:45:51 -05:00
  • 2cc64d71b0 simplify mi350x gemm / viz asm tests (#13984) qazal 2026-01-03 11:11:07 +09:00
  • 7cbafb2ef1 update hypothesis min version (#13983) chenyu 2026-01-02 21:01:57 -05:00
  • 0e240fb987 Merge branch 'master' into amd_sqtt George Hotz 2026-01-02 20:30:16 -05:00
  • d2c1712e4c more tests George Hotz 2026-01-02 17:29:48 -08:00
  • 96b0ee0966 lil George Hotz 2026-01-02 16:53:31 -08:00
  • 9b5c4bc698 shorter George Hotz 2026-01-02 16:48:26 -08:00
  • 6ea3586101 short George Hotz 2026-01-02 16:45:34 -08:00
  • 92cb8b6776 tests pass George Hotz 2026-01-02 16:43:03 -08:00
  • c416b20668 failures George Hotz 2026-01-02 15:54:02 -08:00
  • 415b83ba18 tests pass George Hotz 2026-01-02 15:47:39 -08:00
  • 8c7eacea59 getting close George Hotz 2026-01-02 15:25:18 -08:00
  • 81542699f8 work George Hotz 2026-01-02 14:39:52 -08:00
  • 9dc524536f IMAGE=1 creates "dynamic" images (#13769) Christopher Milan 2026-01-02 13:22:39 -08:00
  • 79f55a5d5e test_snop is correct George Hotz 2026-01-02 12:01:08 -08:00
  • 37518fb236 start with nop George Hotz 2026-01-02 11:40:08 -08:00
  • 672008ccab framework George Hotz 2026-01-02 11:31:41 -08:00
  • 849af761a4 simpler George Hotz 2026-01-02 11:10:40 -08:00
  • 61dc70f1a8 add driving_vision IMAGE=1 benchmark (#13979) Christopher Milan 2026-01-02 10:58:27 -08:00
  • ab46b3d8d3 origin/master George Hotz 2026-01-02 10:47:00 -08:00
  • df20197bfb rever emu to master George Hotz 2026-01-02 10:46:46 -08:00
  • 2b56c264d5 compare tests George Hotz 2026-01-02 10:39:07 -08:00
  • c7e5c2f996 Merge origin/master, remove deleted test_emu.py George Hotz 2026-01-02 09:41:34 -08:00
  • 0e282025ff assembly/amd: split test_emu into hw tests (#13966) George Hotz 2026-01-02 11:04:56 -05:00
  • 2e2b5fed12 fix misspellings (#13976) chenyu 2026-01-02 10:37:38 -05:00
  • f49e4714af Fix spelling errors in README for AMD assembly (#13975) nietras 2026-01-02 16:15:20 +01:00
  • a78fcc55a4 amd tc 1616128 (#13439) b1tg 2026-01-02 22:01:05 +08:00
  • fcbb896e05 remove unused to_struct [pr] (#13973) chenyu 2026-01-02 08:54:57 -05:00
  • 659aa14043 orks George Hotz 2026-01-02 05:29:48 -08:00
  • ff7853a65a am: fix aid doorbells (#13971) nimlgen 2026-01-02 15:53:44 +03:00
  • 42abb0586c am: fix aid doorbells (#13972) nimlgen 2026-01-02 15:53:13 +03:00
  • ebbaad6bfd am: enable all sdma engines (#13970) nimlgen 2026-01-02 15:25:15 +03:00
  • 5f52266225 mi350x gemm: use Tensor.custom_kernel in asm test (#13969) qazal 2026-01-02 18:30:50 +09:00
  • 21ffa1a86b 64 nops George Hotz 2026-01-02 00:38:27 -05:00
  • 29f3fb7af3 still stable George Hotz 2026-01-01 23:45:19 -05:00
  • 1edc7fc519 stable George Hotz 2026-01-01 23:43:43 -05:00
  • c9a3ac988c cleanest George Hotz 2026-01-01 23:18:19 -05:00
  • 5a1a561e0f assembly/amd: rdna4 autogen (#13967) George Hotz 2026-01-01 23:12:18 -05:00
  • 77d96acbe3 clean George Hotz 2026-01-01 22:59:07 -05:00
  • 660ecf272b work George Hotz 2026-01-01 22:50:50 -05:00
  • 267bbb163e progress George Hotz 2026-01-01 21:11:29 -05:00
  • b27527f05a fix: missed inner tracked range (#13964) wozeparrot 2026-01-01 21:09:57 -05:00
  • ecbac8a338 tk: fa cleanups + causal test (#13963) wozeparrot 2026-01-01 21:05:00 -05:00
  • de29a49ea3 all the ones i can find George Hotz 2026-01-01 20:56:30 -05:00
  • 742e10a572 remove fake ones George Hotz 2026-01-01 20:26:53 -05:00
  • 447fe8907b more George Hotz 2026-01-01 20:22:52 -05:00
  • b0cfcec183 good George Hotz 2026-01-01 20:12:20 -05:00
  • 1726084b2a filt George Hotz 2026-01-01 19:40:43 -05:00
  • af0392efea only set DiskDevice.size if it opens successfully (#13962) chenyu 2026-01-01 19:33:26 -05:00
  • de069a4876 many George Hotz 2026-01-01 19:21:46 -05:00
  • 4573e91e61 more George Hotz 2026-01-01 18:51:31 -05:00
  • 8d43212bc6 assembly/amd: start work on SQTT parsing/emulation George Hotz 2026-01-01 18:40:58 -05:00
  • e036d6df89 properly fix DiskDevice reuse (#13961) chenyu 2026-01-01 18:08:23 -05:00
  • dfb813b760 assembly/amd: add pcode ds ops (#13939) George Hotz 2026-01-01 16:24:13 -05:00
  • a8bea4ec52 remove __all__ George Hotz 2026-01-01 16:14:15 -05:00
  • 388514c5b1 better George Hotz 2026-01-01 16:03:29 -05:00