Commits

Branch selector

 main

User selector

ThomasRaoux

Datepicker

All time

Commit History

Commits on Dec 5, 2024

• [BACKEND] Use an optimized ptx code sequence for fp4 upcasting (#5344)

  Show description for 10552c5
  

  ThomasRaoux

  and

  rawnhenry

  authored
  10552c5

Commits on Dec 4, 2024

• Revert "[AMD][BACKEND] Add gfx950 target definitions." (#5342)

  Show description for 6e24f72
  ThomasRaoux

  authored
  6e24f72

• Revert "[LLVM] Update to llvm/llvm-project@1f20eee6dc36" (#5341)

  Show description for 7f2c56a
  ThomasRaoux

  authored
  7f2c56a

Commits on Nov 29, 2024

• Add tests for 3D local_load local_alloc and relax asserts (#5285)

  Show description for cc89dac
  ThomasRaoux

  authored
  cc89dac

Commits on Nov 26, 2024

• [BACKEND] Fold transpose(splat_const) (#5259)

  Show description for deee78f
  

  ThomasRaoux

  and

  peterbell10

  authored
  deee78f

Commits on Nov 25, 2024

• Allow Layouts to propogate to local_load (#5219) (#5249)

  Show description for 09fcc52
  

  ThomasRaoux

  and

  mbrookhart

  authored
  09fcc52

• [PIPELINER] tweak pipeline heuristic (#5247)

  Show description for 4107453
  ThomasRaoux

  authored
  4107453

Commits on Nov 22, 2024

• Revert "[AMD] Use warp shuffle for MFMA to Dot operand layout conversion (FP8)" (#5240)

  Show description for c6da81a
  ThomasRaoux

  authored
  c6da81a

• Revert "Allow Layouts to propogate to local_load" (#5237)

  Show description for 84ced0e
  ThomasRaoux

  authored
  84ced0e

• [BACKEND] Fix a missed transpose optimization during refactor (#5236)

  ThomasRaoux

  authored
  340cbc6

• [BACKEND] Fix transpose optimization missed during refactor (#5226)

  ThomasRaoux

  authored
  4ae95e7

Commits on Nov 21, 2024

• [AMD] Prevent wrong reordering of scf operations (#5203)

  Show description for c8a31a0
  ThomasRaoux

  authored
  c8a31a0

Commits on Nov 20, 2024

• [MXFP] Fix packing for mxfp4 type (#5197)

  Show description for 7088c64
  ThomasRaoux

  authored
  7088c64

Commits on Nov 19, 2024

• [BACKEND] Fix getElemsPerThread for mmav3 dot operand (#5189)

  Show description for aaf64d6
  ThomasRaoux

  authored
  aaf64d6

Commits on Nov 18, 2024

• [BACKEND] Add missing precondition in optimize acc init (#5184)

  Show description for c76b342
  ThomasRaoux

  authored
  c76b342

Commits on Nov 14, 2024

• [TEST] Make mixed matmul test deterministic (#5151)

  Show description for a9dda9e
  ThomasRaoux

  authored
  a9dda9e

Commits on Nov 13, 2024

• [BACKEND] Support scalar fp_to_fp (#5132)

  Show description for d556ce9
  ThomasRaoux

  authored
  d556ce9

Commits on Nov 11, 2024

• Support scaled_dot with rhs scale (#5107)

  Show description for 462de12
  ThomasRaoux

  authored
  462de12

Commits on Nov 6, 2024

• [BACKEND] Fix reduce with slice layout inputs (#5080)

  Show description for 35f1827
  ThomasRaoux

  authored
  35f1827

Commits on Nov 2, 2024

• Add back barrier after asserts (#5043)

  Show description for 92a4fad
  ThomasRaoux

  authored
  92a4fad

• Revert "[BACKEND] Add barrier after assert op to avoid race condition" (#5047)

  Show description for 6c72dbe
  ThomasRaoux

  authored
  6c72dbe

Commits on Nov 1, 2024

• [BACKEND] Add barrier after assert op to avoid race condition (#5035)

  Show description for d0db12b
  ThomasRaoux

  authored
  d0db12b

Commits on Oct 8, 2024

• revert #4774 (#4873)

  Show description for 9e72047
  ThomasRaoux

  authored
  9e72047

• Revert #4784 (#4865)

  Show description for ca70f08
  ThomasRaoux

  authored
  ca70f08

Commits on Oct 3, 2024

• Fix assert loc for cases where assert is in an inlined func (#4840)

  ThomasRaoux

  authored
  b24fa65

Commits on Sep 21, 2024

• [BACKEND] Switch back to use llvm.load for shared memory load (#4776)

  Show description for 576426b
  ThomasRaoux

  authored
  576426b

• Revert "Use fast math function for tl.math.log as exp (#4723)" (#4779)

  Show description for 3a647f0
  ThomasRaoux

  authored
  3a647f0

• Turn off fast log (#4777)

  Show description for 20f4b41
  ThomasRaoux

  authored
  20f4b41

Commits on Sep 13, 2024

• Use fast math function for tl.math.log as exp (#4723)

  Show description for 84fe9da
  ThomasRaoux

  authored
  84fe9da

Commits on Sep 11, 2024

• [BACKEND] Don't use MMAV3 when K dim is smaller than native size (#4700)

  Show description for 24f8dbd
  ThomasRaoux

  authored
  24f8dbd

Commits on Sep 10, 2024

• [BACKEND] Turn off thread locality optimization based on assumptions (#4688)

  Show description for 58eccfc
  ThomasRaoux

  authored
  58eccfc

Commits on Sep 9, 2024

• [BACKEND] Fix potential bug in membar TMA rules (#4681)

  Show description for a63a477
  ThomasRaoux

  authored
  a63a477

• [BACKEND] Allow backend to specify special rules for membar insertion (#4675)

  Show description for 8a3fb7e
  ThomasRaoux

  authored
  8a3fb7e

Commits on Sep 5, 2024

• [BACKEND] Relax layout supported by SplitOp (#4653)

  Show description for 2f6cec9
  ThomasRaoux

  authored
  2f6cec9

• [BACKEND] Reduce shared memory usage when pipelining multiple TMA stores (#4654)

  Show description for 9112cfa
  ThomasRaoux

  authored
  9112cfa

Pagination

Previous

Next