[llvm] [AArch64][Machine-Combiner] Split loads into lanes of neon vectors into multiple vectors when possible (PR #142941)

[Jonathan Cohen via llvm-commits]

================
@@ -7327,6 +7329,9 @@ bool AArch64InstrInfo::isThroughputPattern(unsigned Pattern) const {
   case AArch64MachineCombinerPattern::MULSUBv2i32_indexed_OP2:
   case AArch64MachineCombinerPattern::MULSUBv4i32_indexed_OP1:
   case AArch64MachineCombinerPattern::MULSUBv4i32_indexed_OP2:
+  case AArch64MachineCombinerPattern::GATHER_i32:
----------------
jcohen-apple wrote:

Done

https://github.com/llvm/llvm-project/pull/142941