[all-commits] [llvm/llvm-project] 4d2083: [WebAssembly] Support f16x8.demote_f32x4_zero (#19...

[Brendan Dahl via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 4d20831a6562ecc5e0b87d5785b2911060c3ab41
      https://github.com/llvm/llvm-project/commit/4d20831a6562ecc5e0b87d5785b2911060c3ab41
  Author: Brendan Dahl <brendan.dahl at gmail.com>
  Date:   2026-04-27 (Mon, 27 Apr 2026)

  Changed paths:
    M llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
    M llvm/lib/Target/WebAssembly/WebAssemblyInstrSIMD.td
    M llvm/test/CodeGen/WebAssembly/f16-intrinsics.ll

  Log Message:
  -----------
  [WebAssembly] Support f16x8.demote_f32x4_zero (#193564)

Add support for the f16x8.demote_f32x4_zero instruction. This
instruction converts a v4f32 vector to a v4f16 and pads the result with
zeros to fill the 128-bit register.

This enables efficient lowering of fptrunc operations from v4f32 to
v4f16 when the result is zero-extended or when only the low lanes are
needed. A DAG combine is included to recognize these patterns and fold
them into the new instruction.



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