[all-commits] [llvm/llvm-project] 9567f4: [AMDGPU][GlobalISel] Use single B32 rule for perml...

[Chinmay Deshpande via All-commits]

  Branch: refs/heads/users/chinmaydd/permlanee
  Home:   https://github.com/llvm/llvm-project
  Commit: 9567f477d7e0b72df82d5194d2f96834200545b0
      https://github.com/llvm/llvm-project/commit/9567f477d7e0b72df82d5194d2f96834200545b0
  Author: Chinmay Deshpande <chdeshpa at amd.com>
  Date:   2026-08-04 (Tue, 04 Aug 2026)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp

  Log Message:
  -----------
  [AMDGPU][GlobalISel] Use single B32 rule for permlane16

B32 already covers s32, <2 x s16> and 32-bit pointers, so one StandardB
fast rule replaces the separate S32, V2S16 and DivPtr32 rules. This
matches permlane_bcast/up/down/xor and permlane64, which have the same
overloaded intrinsic shape.

Change-Id: I770a93d5e75e68df704e84d7ad165fdd2cf2fc53



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