[all-commits] [llvm/llvm-project] 1a219e: [X86] Support EVEX compression from MOVBErr to BSW...

[XinWang10 via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 1a219e989f07eef06fbc9f75ad0c07ebdc652334
      https://github.com/llvm/llvm-project/commit/1a219e989f07eef06fbc9f75ad0c07ebdc652334
  Author: XinWang10 <108658776+XinWang10 at users.noreply.github.com>
  Date:   2024-01-30 (Tue, 30 Jan 2024)

  Changed paths:
    M llvm/lib/Target/X86/X86CompressEVEX.cpp
    M llvm/test/CodeGen/X86/apx/compress-evex.mir
    M llvm/utils/TableGen/X86ManualCompressEVEXTables.def

  Log Message:
  -----------
  [X86] Support EVEX compression from MOVBErr to BSWAP (#79775)

APX promoted MOVBE instructions were supported in #77431. The reg2reg
variants of MOVBE are newly introduced by APX and can be optimized to
BSWAP instruction when the 2 register operands are same.

This patch adds manual entries for MOVBErr instructions when we do ndd
to non-ndd compression #77731.
RFC:
https://discourse.llvm.org/t/rfc-design-for-apx-feature-egpr-and-ndd-support/73031/4