[PATCH] D117219: [LSV] Vectorize loads of vectors by turning it into a larger vector

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added a comment.

Thanks, I completely forgot this didn't handle this



================
Comment at: llvm/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp:1303
+  // Bitcast might not be an Instruction, if the value being loaded is a
+  // constant.  In that case, no need to reorder anything.
+  if (Instruction *BitcastInst = dyn_cast<Instruction>(Bitcast))
----------------
Too many spaces before In?


================
Comment at: llvm/test/Transforms/LoadStoreVectorizer/NVPTX/4x2xhalf.ll:5
+entry:
+  %0 = bitcast half* %rd0 to <2 x half>*
+  %1 = load <2 x half>, <2 x half>* %0, align 16
----------------
Should use named values in tests


================
Comment at: llvm/test/Transforms/LoadStoreVectorizer/NVPTX/4x2xhalf.ll:33
+  ret void
+
+; CHECK-LABEL: @ldg_f16
----------------
Should add a larger variety of tests with more vector widths, particularly some 3 element vectors


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D117219/new/

https://reviews.llvm.org/D117219