[PATCH] D74581: [llvm][CodeGen][aarch64] Add continuous prefetch intrinsics for SVE.

[Sander de Smalen via Phabricator via llvm-commits]

sdesmalen accepted this revision.
sdesmalen added a comment.
This revision is now accepted and ready to land.

nit: in the title:
s/continuous/contiguous/

LGTM otherwise



================
Comment at: llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp:4414
 
+namespace {
+/// When \p PredVT is a scalable vector predicate in the form
----------------
nit: Any reason you're using an anonymous namespace over `static` for these functions?


================
Comment at: llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp:4424
+
+  if (PredVT.getVectorElementType() != MVT::i1)
+    return EVT();
----------------
nit: you can fold this into the condition above.


================
Comment at: llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp:4432
+
+  EVT ScalarVT = EVT::getIntegerVT(Ctx, 128 / NumElts);
+  EVT MemVT = EVT::getVectorVT(Ctx, ScalarVT, NumElts, /*IsScalable=*/true);
----------------
instead of `128`, can you use `AArch64::SVEBitsPerBlock`?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D74581/new/

https://reviews.llvm.org/D74581