[all-commits] [llvm/llvm-project] 58950d: [mlir][nvvm] Implement `mbarrier.init`

[Guray Ozen via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 58950d4addd6d1dd920801b32cc75ddc8b9f6c3a
      https://github.com/llvm/llvm-project/commit/58950d4addd6d1dd920801b32cc75ddc8b9f6c3a
  Author: Guray Ozen <guray.ozen at gmail.com>
  Date:   2023-06-16 (Fri, 16 Jun 2023)

  Changed paths:
    M mlir/include/mlir/Dialect/LLVMIR/NVVMOps.td
    M mlir/test/Dialect/LLVMIR/nvvm.mlir

  Log Message:
  -----------
  [mlir][nvvm] Implement `mbarrier.init`

NV GPUs provides split arrive/wait barriers that one can syncronize a subgroup of threads in CTA. It is particularly important for Hopper GPUs and allows tracking engines like TMA. See for more details:
https://docs.nvidia.com/cuda/parallel-thread-execution/#parallel-synchronization-and-communication-instructions-mbarrier

This initial implementation sets the foundation for future enhancements and additions.

Reviewed By: qcolombet

Differential Revision: https://reviews.llvm.org/D151334