[all-commits] [llvm/llvm-project] c507a0: [NVPTX] Add TMA Bulk Copy Intrinsics (#138679)

[Durgadoss R via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: c507a0830df2e4fd0c234eee035aac2109de6d6e
      https://github.com/llvm/llvm-project/commit/c507a0830df2e4fd0c234eee035aac2109de6d6e
  Author: Durgadoss R <durgadossr at nvidia.com>
  Date:   2025-05-15 (Thu, 15 May 2025)

  Changed paths:
    M llvm/docs/NVPTXUsage.rst
    M llvm/include/llvm/IR/IntrinsicsNVVM.td
    M llvm/lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp
    M llvm/lib/Target/NVPTX/NVPTXISelDAGToDAG.h
    M llvm/lib/Target/NVPTX/NVPTXIntrinsics.td
    A llvm/test/CodeGen/NVPTX/cp-async-bulk-s2g-sm100.ll
    M llvm/test/CodeGen/NVPTX/cp-async-bulk.ll

  Log Message:
  -----------
  [NVPTX] Add TMA Bulk Copy Intrinsics (#138679)

This patch adds a new variant of TMA Bulk Copy
intrinsics introduced in sm100+. This variant
has an additional byte_mask to select the bytes
for the copy operation.

* Selection is all done through table-gen now.
  So, this patch removes the corresponding
  SelectCpAsyncBulkS2G() function.
* lit tests are verified with a cuda-12.8 ptxas
  executable.

PTX Spec link:

https://docs.nvidia.com/cuda/parallel-thread-execution/#data-movement-and-conversion-instructions-bulk-copy

Signed-off-by: Durgadoss R <durgadossr at nvidia.com>



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications