[all-commits] [llvm/llvm-project] 778dd2: [AArch64][NeoverseV1] Load/Store Register Half wit...

Julien Villette via All-commits all-commits at lists.llvm.org
Thu Oct 1 01:41:20 PDT 2026


  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 778dd2c1b8050f69bee779893c27e7647c07eff3
      https://github.com/llvm/llvm-project/commit/778dd2c1b8050f69bee779893c27e7647c07eff3
  Author: Julien Villette <julien.villette at gmail.com>
  Date:   2026-10-01 (Thu, 01 Oct 2026)

  Changed paths:
    M llvm/lib/Target/AArch64/AArch64SchedNeoverseV1.td
    M llvm/lib/Target/AArch64/AArch64SchedPredNeoverse.td
    M llvm/test/tools/llvm-mca/AArch64/Neoverse/V1-basic-instructions.s

  Log Message:
  -----------
  [AArch64][NeoverseV1] Load/Store Register Half with scaling (#222282)

On Neoverse V1, LDRH/STRH with scaling has a specific execution compared
to other LDR/STR.

Opcode            | Latency   | Pipelines
-------------------------------------------
Common LDR        | 4         | L
LDRH with scaling | 5         | I, L
Common STR        | 1         | L01, D
STRH with scaling | 2         | I, L01, D

Verified on Neoverse V1 with micro benchmarks and compared all
addressing mode variantes. Neoverse V2 is unaffected (confirmed by SOG
documentation and microbenchmarks).



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


More information about the All-commits mailing list