Code

ROCm Sparse Marlin Kernels#1206

Code

Open

petrex wants to merge 32 commits intopytorch:mainpytorch/ao:mainfrom petrex:rocm_sparse_marlinpetrex/ao:rocm_sparse_marlinCopy head branch name to clipboard

+219-16

Commits

Commits on Oct 16, 2024

• enable build for rocm for fp6_llm

  lcskrishna

  committed
  6d92e40

  Copy full SHA for 6d92e40

Commits on Oct 17, 2024

• Merge pull request #1 from lcskrishna/cl/rocm-enablement

  Show description for 14b3fce
  petrex

  authored
  14b3fce

  Copy full SHA for 14b3fce

Commits on Oct 23, 2024

• enable tiled layout extension

  lcskrishna

  committed
  f1a22cf

  Copy full SHA for f1a22cf

• fix build error related to option

  petrex

  committed
  0bef6ca

  Copy full SHA for 0bef6ca

• require rocm 6.2

  petrex

  committed
  893ae03

  Copy full SHA for 893ae03

Commits on Oct 24, 2024

• enable tensor tiled layout extension with successful compilation

  lcskrishna

  committed
  a0d3788

  Copy full SHA for a0d3788

• enable successful build

  lcskrishna

  committed
  e4e654d

  Copy full SHA for e4e654d

Commits on Oct 29, 2024

• clean-up

  petrex

  committed
  3e2c6a1

  Copy full SHA for 3e2c6a1

• Merge pull request #3 from lcskrishna/csrikris_enable_tensor_tile

  Show description for c86880e
  petrex

  authored
  c86880e

  Copy full SHA for c86880e

• fix potential memory access issue

  petrex

  committed
  91d3c75

  Copy full SHA for 91d3c75

Commits on Nov 12, 2024

• fix __nv_bfloat162 init

  petrex

  committed
  38b7d1c

  Copy full SHA for 38b7d1c

• add comment for MI300x isa

  petrex

  committed
  279f4b3

  Copy full SHA for 279f4b3

Commits on Nov 18, 2024

• Merge branch 'main' into rocm_enablement_staging

  petrex

  authored
  612ad14

  Copy full SHA for 612ad14

Commits on Jan 6, 2025

• fix build for non-rocm

  lcskrishna

  committed
  bbf5a72

  Copy full SHA for bbf5a72

• Merge pull request #4 from lcskrishna/rocm_enablement

  Show description for 735570e
  petrex

  authored
  735570e

  Copy full SHA for 735570e

• Merge branch 'main' into rocm_enablement_staging

  petrex

  authored
  253c188

  Copy full SHA for 253c188

• add sparse_marlin kernel to the build

  petrex

  committed
  a2f1736

  Copy full SHA for a2f1736

• drop .h from conversion

  petrex

  committed
  f817edf

  Copy full SHA for f817edf

• cp_asyc4_pred_zfill() AMD implementation

  petrex

  committed
  c9bc1bc

  Copy full SHA for c9bc1bc

• implement matching mem utility with amd GCN isa

  petrex

  committed
  16feff4

  Copy full SHA for 16feff4

• implement mma util with amd gcn isa

  petrex

  committed
  0b21555

  Copy full SHA for 0b21555

• enable rocm path

  petrex

  committed
  f23b194

  Copy full SHA for f23b194

• update copy from global to lds

  

  lcskrishna

  authored and

  petrex

  committed
  ecc3927

  Copy full SHA for ecc3927

• implement cvta_to_shared()

  petrex

  committed
  a80730b

  Copy full SHA for a80730b

• consolidate code with cvta_to_shared()

  petrex

  committed
  d2c7ce4

  Copy full SHA for d2c7ce4

Commits on Jan 8, 2025

• Merge branch 'main' into rocm_sparse_marlin

  petrex

  authored
  15974c7

  Copy full SHA for 15974c7

• lint

  petrex

  committed
  a4e8c30

  Copy full SHA for a4e8c30

Commits on Jan 9, 2025

• add GPU arch check for MI300x

  petrex

  committed
  c678cb0

  Copy full SHA for c678cb0

• revert change in tensor_core_tile_layout.cu

  petrex

  committed
  08d1cfb

  Copy full SHA for 08d1cfb

Commits on Jan 15, 2025

• Merge branch 'main' into rocm_sparse_marlin

  petrex

  authored
  b96196b

  Copy full SHA for b96196b

• lint

  Show description for aea9d81
  petrex

  committed
  aea9d81

  Copy full SHA for aea9d81

Commits on Feb 25, 2025

• Merge branch 'main' into rocm_sparse_marlin

  petrex

  authored
  f18043d

  Copy full SHA for f18043d