llvm-project

04a26a03 - [X86] Add ISD::VECREDUCE_MUL lowering instead of matching in DAG (#207593)

[X86] Add ISD::VECREDUCE_MUL lowering instead of matching in DAG (#207593) Add X86 support for ISD::VECREDUCE_MUL and remove it from combineArithReduction - most of the change is refactoring the custom promotion of vXi8 -> vXi16. Another part of #194621