Commits

Branch selector

 develop

User selector

All users

Datepicker

All time

Commit History

Commits on May 20, 2025

• added gemm universal example in readme (#2216)

  AviralGoelAMD

  authored
  990d645

  Copy full SHA for 990d645

• Fix 11_add_rmsnorm2d_rdquant (#2207)

  SamiAario-AMD

  authored
  380bca2

  Copy full SHA for 380bca2

• Add the instances for small sized GEMM in preshuffle and improve CMake Flag (#2212)

  Show description for 1386924
  ThomasNing

  authored
  1386924

  Copy full SHA for 1386924

• [CK_TILE] Grouped GEMM tile loop (#2146)

  Show description for d1e6f09
  

  samremes

  and

  aosewski

  authored
  d1e6f09

  Copy full SHA for d1e6f09

• remove debug statements from CMakeLists (#2204)

  AviralGoelAMD

  authored
  c492922

  Copy full SHA for c492922

• [CMake] Disable newly added compiler warning -Wnrvo (#2210)

  Show description for 0970f22
  jplehr

  authored
  0970f22

  Copy full SHA for 0970f22

• Use new mfma instructions for FP8 on gfx950 (#2202)

  Show description for f181700
  

  jefyang1

  and

  andriy-ca

  authored
  f181700

  Copy full SHA for f181700

Commits on May 19, 2025

• MX GEMM - Expand MX MFMA Testing to BF8, FP6, and BF6 Data Types (#2199)

  Show description for 57e0f5d
  

  andriy-ca

  and

  geyyer

  authored
  57e0f5d

  Copy full SHA for 57e0f5d

• Fix example_grouped_gemm_multiple_d_xdl_fp16 on gfx950 (#2203)

  Show description for b8b12bb
  jefyang1

  authored
  b8b12bb

  Copy full SHA for b8b12bb

Commits on May 17, 2025

• Restore oddc instances (#2201)

  bartekxk

  authored
  6342f6b

  Copy full SHA for 6342f6b

Commits on May 16, 2025

• Narrowing error fix for codegen compilation (#2194)

  Show description for 5b3430b
  

  arai713

  and

  ThomasNing

  authored
  5b3430b

  Copy full SHA for 5b3430b

• Build and store CK library deb package for all targets daily. (#2196)

  Show description for 40668c9
  illsilin

  authored
  40668c9

  Copy full SHA for 40668c9

• Disable conv for Filter1x1Stride1Pad0 when K or C is even (#2186)

  mozga-amd

  authored
  fa3c681

  Copy full SHA for fa3c681

• [CK_TILE] fMHA batch_prefill block index & logits soft-capping optimizations (#2198)

  Show description for 791802b
  poyenc

  authored
  791802b

  Copy full SHA for 791802b

Commits on May 15, 2025

• Use only qr_async pipeline for batch_prefill (#2195)

  poyenc

  authored
  8cb0474

  Copy full SHA for 8cb0474

• Adding validation for tile sizes in Tile Engine (#2189)

  Show description for 3d8d6e7
  amd-khushbu

  authored
  3d8d6e7

  Copy full SHA for 3d8d6e7

• Extend 64x64 with 4 waves instances for grouped conv bwd wei (#2187)

  Show description for 7c0e29c
  bartekxk

  authored
  7c0e29c

  Copy full SHA for 7c0e29c

Commits on May 14, 2025

• fix moe sorting build fail (#2190)

  Show description for 41c17d0
  

  solinzby1

  and

  solinzby

  authored
  41c17d0

  Copy full SHA for 41c17d0

Commits on May 13, 2025

• Update the buffer load/store intrinsic names for clang>=20. (#2192)

  Show description for 58f9e9f
  illsilin

  authored
  58f9e9f

  Copy full SHA for 58f9e9f

• Switch to v2 pipeline for grouped conv bwd data (#2181)

  Show description for c53b7bd
  bartekxk

  authored
  c53b7bd

  Copy full SHA for c53b7bd

• [CK_TILE] Add logits soft-capping & customization support to the FMHA forward kernel/pipelines (#2163)

  Show description for 2920604
  

  3 peopleauthored
  2920604

  Copy full SHA for 2920604

Commits on May 12, 2025

• Disable SMFMA gfx90a (#2184)

  Show description for f05e45b
  amd-khushbu

  authored
  f05e45b

  Copy full SHA for f05e45b

• Improve the general performance of the Preshuffled GEMM V3 & delete the unnecessary instances (#2166)

  Show description for b49f7de
  ThomasNing

  authored
  b49f7de

  Copy full SHA for b49f7de

• Vectorized Transpose for Batched Transpose CK Tile Operator (#2131)

  Show description for 9d1e44e
  

  ThomasNing

  and

  ThruptiRajLakshmanaGowda

  authored
  9d1e44e

  Copy full SHA for 9d1e44e

Commits on May 11, 2025

• Support for swizzle and transpose for MFMA_16x16x32_F16/BF16 (#2172)

  Show description for d8faf1c
  amd-khushbu

  authored
  d8faf1c

  Copy full SHA for d8faf1c

Commits on May 9, 2025

• Add grouped conv fwd bias relu instances (#2179)

  Show description for 6fddb57
  bartekxk

  authored
  6fddb57

  Copy full SHA for 6fddb57

• Fix grouped conv bwd data tests on gfx950 (#2173)

  jefyang1

  authored
  6b1a339

  Copy full SHA for 6b1a339

• fix moe gemm2 for gfx950 (#2164)

  Show description for a233901
  

  mtgu0705

  and

  mtgu0705

  authored
  a233901

  Copy full SHA for a233901

• Disable SMFMA for gfx90a (#2182)

  amd-khushbu

  authored
  ef72a4b

  Copy full SHA for ef72a4b

Commits on May 8, 2025

• Generate ckProfiler package for gfx942 only. (#2180)

  Show description for 3448e12
  illsilin

  authored
  3448e12

  Copy full SHA for 3448e12

• Ensure MX GEMM Instances can be Cross-Compiled for Multiple Architectures (#2171)

  Show description for cb27e7c
  andriy-ca

  authored
  cb27e7c

  Copy full SHA for cb27e7c

• Revert "Disable the SMFMA instruction for gfx90a. (#2174)" (#2175)

  Show description for c757046
  ThomasNing

  authored
  c757046

  Copy full SHA for c757046

• Disable the SMFMA instruction for gfx90a. (#2174)

  Show description for a32d907
  amd-khushbu

  authored
  a32d907

  Copy full SHA for a32d907

• Flatmm merge (#2168)

  Show description for 6a3960c
  

  solinzby1

  and

  solinzby

  authored
  6a3960c

  Copy full SHA for 6a3960c

• [CK_Tile] Simplified Mem pipeline (#2159)

  Show description for c7b8e86
  

  amd-khushbu

  and

  ThomasNing

  authored
  c7b8e86

  Copy full SHA for c7b8e86

Pagination

Previous

Next