Code

Support TMA with 64-bit indexing#3599

Code

Open

jacobhinkle wants to merge 10 commits intomainNVIDIA/Fuser:mainfrom tma_int64_indexingNVIDIA/Fuser:tma_int64_indexingCopy head branch name to clipboard

+62-10

Commits

Commits on Dec 16, 2024

• Support TMA with 64-bit indexing

  jacobhinkle

  committed
  2e66b80

  Copy full SHA for 2e66b80

Commits on Dec 17, 2024

• Merge remote-tracking branch 'origin/main' into tma_int64_indexing

  jacobhinkle

  committed
  ade644f

  Copy full SHA for ade644f

• Add test

  jacobhinkle

  committed
  5ec3ee5

  Copy full SHA for 5ec3ee5

• Cast to int32_t instead of uint32_t

  jacobhinkle

  committed
  308d4fe

  Copy full SHA for 308d4fe

• Add check for TMA axes when using 64-bit indexing

  jacobhinkle

  committed
  1df92f9

  Copy full SHA for 1df92f9

• Fix type of literal that caused overflow in check

  jacobhinkle

  committed
  9bdf4c8

  Copy full SHA for 9bdf4c8

• Cast in the IR instead of the runtime memory.cu

  jacobhinkle

  committed
  d1c71c6

  Copy full SHA for d1c71c6

• Fix inequality in check

  jacobhinkle

  committed
  8e55b54

  Copy full SHA for 8e55b54

Commits on Dec 26, 2024

• clang-tidy

  jacobhinkle

  committed
  be3b2cd

  Copy full SHA for be3b2cd

• Merge branch 'main' into tma_int64_indexing

  jacobhinkle

  authored
  3db6cc2

  Copy full SHA for 3db6cc2