[all-commits] [llvm/llvm-project] 7487c7: [AMDGPU] Change 9 SWMMAC builtins to use 64-bit in...

[Stanislav Mekhanoshin via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 7487c7581ea288d004328ae385bf60675a0d6892
      https://github.com/llvm/llvm-project/commit/7487c7581ea288d004328ae385bf60675a0d6892
  Author: Stanislav Mekhanoshin <Stanislav.Mekhanoshin at amd.com>
  Date:   2026-02-12 (Thu, 12 Feb 2026)

  Changed paths:
    M clang/include/clang/Basic/BuiltinsAMDGPU.td
    A clang/test/CodeGenOpenCL/.gdb_history
    M clang/test/CodeGenOpenCL/builtins-amdgcn-gfx1250-wmma-w32.cl
    M llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
    M llvm/lib/Target/AMDGPU/SIISelLowering.cpp
    M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.wmma.index.gfx1250.w32.ll

  Log Message:
  -----------
  [AMDGPU] Change 9 SWMMAC builtins to use 64-bit index (#181246)

There 9 gfx1250 instructions have 64-bit packed index:

- v_swmmac_f16_16x16x128_bf8_bf8
- v_swmmac_f16_16x16x128_bf8_fp8
- v_swmmac_f16_16x16x128_fp8_bf8
- v_swmmac_f16_16x16x128_fp8_fp8
- v_swmmac_f32_16x16x128_bf8_bf8
- v_swmmac_f32_16x16x128_bf8_fp8
- v_swmmac_f32_16x16x128_fp8_bf8
- v_swmmac_f32_16x16x128_fp8_fp8
- v_swmmac_i32_16x16x128_iu8

Intrinsics accept anyint, but builtins are defined with i32 argument.

Fixes: SWDEV-579843



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