[all-commits] [llvm/llvm-project] 865952: [NVPTX] Add conversion intrinsics from/to fp8 type...

[Sergey Kozub via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 865952bbbb97655410d2c34c21938134059c4076
      https://github.com/llvm/llvm-project/commit/865952bbbb97655410d2c34c21938134059c4076
  Author: Sergey Kozub <skozub at nvidia.com>
  Date:   2024-08-19 (Mon, 19 Aug 2024)

  Changed paths:
    M clang/include/clang/Basic/BuiltinsNVPTX.def
    M clang/test/CodeGen/builtins-nvptx.c
    M llvm/include/llvm/IR/IntrinsicsNVVM.td
    M llvm/lib/Target/NVPTX/NVPTXInstrInfo.td
    M llvm/lib/Target/NVPTX/NVPTXIntrinsics.td
    A llvm/test/CodeGen/NVPTX/convert-sm89.ll

  Log Message:
  -----------
  [NVPTX] Add conversion intrinsics from/to fp8 types (e4m3, e5m2) (#102969)

PTX ISA 8.1 supports FP8 conversions:

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

This PR adds the support for:

- cvt.rn.satfinite{.relu}.f8x2type.f32 d, a, b;
- cvt.rn.satfinite{.relu}.f8x2type.f16x2 d, a;
- cvt.rn.{.relu}.f16x2.f8x2type d, a;

where .f8x2type = { .e4m3x2, .e5m2x2 };



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