[all-commits] [llvm/llvm-project] 5d2ece: [X86] Allow znver3/4/5 targets to use double-shift...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 5d2ece14378ef00b569e5f29034a0236d735da44
      https://github.com/llvm/llvm-project/commit/5d2ece14378ef00b569e5f29034a0236d735da44
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2025-06-25 (Wed, 25 Jun 2025)

  Changed paths:
    M llvm/lib/Target/X86/X86.td
    M llvm/test/CodeGen/X86/shift-i512.ll
    M llvm/test/CodeGen/X86/x86-64-double-shifts-var.ll

  Log Message:
  -----------
  [X86] Allow znver3/4/5 targets to use double-shift instructions by default (#132720)

While still not as fast as Intel targets, recent AMD znver3 + later CPUs
are not as microcoded+bottlenecked as previous AMD targets (now only
~2cy rthroughput) which improves on the expanded 3*shift+not+or sequence
we expand to as an alternative.

Noticed while triaging #132601



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