[llvm] [CodeGen][PreISelIntrinsicLowering] Add VP-based lowering for memcpy/memmove/memset (PR #165585)

[Luke Lau via llvm-commits]

lukel97 wrote:

This might be a better fit for RISC-V where AFAIK we don't have any vectorized implementations of these lib functions. Do you have a RISC-V version of the patch available? I think this could be an interesting area to explore. 

It would be good to know what the effects on code size and performance are on some benchmarks e.g. llvm-test-suite or SPEC. Are you able to collect some numbers on those? 


https://github.com/llvm/llvm-project/pull/165585