[all-commits] [llvm/llvm-project] f4db03: [AMDGPU] In `LowerDYNAMIC_STACKALLOC`, hoist the r...

[Juan Manuel Martinez Caamaño via All-commits]

  Branch: refs/heads/users/jmmartinez/dynamic_stack_up
  Home:   https://github.com/llvm/llvm-project
  Commit: f4db03884d11c4a5e875632cf33acd77226034da
      https://github.com/llvm/llvm-project/commit/f4db03884d11c4a5e875632cf33acd77226034da
  Author: Juan Manuel Martinez Caamaño <jmartinezcaamao at gmail.com>
  Date:   2026-06-04 (Thu, 04 Jun 2026)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIISelLowering.cpp
    M llvm/test/CodeGen/AMDGPU/amdgpu-cs-chain-fp-nosave.ll
    M llvm/test/CodeGen/AMDGPU/dynamic_stackalloc.ll
    M llvm/test/CodeGen/AMDGPU/llvm.sponentry.ll

  Log Message:
  -----------
  [AMDGPU] In `LowerDYNAMIC_STACKALLOC`, hoist the readfirstlane up one instruction

Instead of:

```
$max_size_vgpr = wave_reduction_umax($vgpr_alloca_size)
$sgpr_newsp = readfirstlane($max_size_vgpr + $sgpr_sp)
```

Hoist the readfirstlane up to perform the addition using scalar
registers:

```
$max_size_sgpr = readfirstlane(wave_reduction_umax($vgpr_alloca_size))
$sgpr_newsp = $max_size_sgpr + $sgpr_sp
```



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