[all-commits] [llvm/llvm-project] 5208f7: [AMDGPU] Fix SIFoldOperandsImpl::canUseImmWithOpSe...

[Daniil Fukalov via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 5208f722d80c9033ed9b5de86e6302f452f4ceca
      https://github.com/llvm/llvm-project/commit/5208f722d80c9033ed9b5de86e6302f452f4ceca
  Author: Daniil Fukalov <dfukalov at gmail.com>
  Date:   2025-05-30 (Fri, 30 May 2025)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIFoldOperands.cpp
    M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cvt.scalef32.pk.gfx950.ll

  Log Message:
  -----------
  [AMDGPU] Fix SIFoldOperandsImpl::canUseImmWithOpSel() for VOP3 packed [B]F16 imms. (#142142)

VOP3 instructions ignore opsel source modifiers, so a constant that
contains two different [B]F16 imms cannot be encoded into instruction
with an src opsel.

E.g. without the fix the following instructions

`s_mov_b32 s0, 0x40003c00 // <half 1.0, half 2.0>`
`v_cvt_scalef32_pk_fp8_f16 v0, s0, v2`

lose `2.0` imm and are folded into

`v_cvt_scalef32_pk_fp8_f16 v1, 1.0, 1.0`

Fixes SWDEV-531672



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