[PATCH] D71773: [AArch64][SVE] Update the definition of AdvSIMD_GatherLoad_VecTorBase_Intrinsic

[Eli Friedman via Phabricator via llvm-commits]

efriedma added a comment.

> better reflect the difference in the semantics of the corresponding instructions

You could make the argument that the base in int_aarch64_sve_ld1_gather is a pointer, and we should semantically require that all the loaded values should point into the same memory allocation.  That would be a significant semantic difference.  (The ACLE distinguishes between "scalar base, vector offset in bytes" and "vector base, scalar offset in bytes".) If that's the distinction we want to make, though, we shouldn't add ImmArg to int_aarch64_sve_ld1_gather_imm; instead, we should rename it to int_aarch64_sve_ld1_gather_scalaroffset, and add patterns for a non-immediate scalar offset.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D71773/new/

https://reviews.llvm.org/D71773