[llvm] [AArch64] Remove extra fmov after half-vector Neon LD instruction (PR #216758)
Jonathan Thackray via llvm-commits
llvm-commits at lists.llvm.org
Mon Aug 17 10:13:13 PDT 2026
================
@@ -700,6 +700,37 @@ bool AArch64MIPeepholeOptImpl::visitINSviGPR(MachineInstr &MI, unsigned Opc) {
return true;
}
+static bool isZeroingDRegLoadOpcode(unsigned int Op) {
+ switch (Op) {
+ case AArch64::LD1Twov1d:
+ case AArch64::LD1Twov1d_POST:
----------------
jthackray wrote:
Might be worth considering adding a flag in tablegen if this list gets too long, and then check it from C++, instead of long case statement.
https://github.com/llvm/llvm-project/pull/216758
More information about the llvm-commits
mailing list