[PATCH] D86225: [AArch64] Optimize instruction selection for certain vector shuffles

[Dave Green via Phabricator via llvm-commits]

dmgreen added a comment.

Looks nice. Perhaps add a few more negative tests for the various different "return false"'s



================
Comment at: llvm/lib/Target/AArch64/AArch64ISelLowering.cpp:7392-7394
+        continue;
+      }
+      if (BlockElts[I] != Elt)
----------------
You can make this an else and drop the continue.


================
Comment at: llvm/lib/Target/AArch64/AArch64ISelLowering.cpp:7407
+                         "previous cases, e.g. isSplat()");
+  if (!FirstRealElt) {
+    DupLaneOp = 0;
----------------
Would this be nullptr or BlockElts.end()?


================
Comment at: llvm/lib/Target/AArch64/AArch64ISelLowering.cpp:7425
+  // undefs)
+  for (size_t I = 0; I < BlockElts.size(); I++)
+    if (BlockElts[I] >= 0 && (unsigned)BlockElts[I] != Elt0 + I)
----------------
BlockElts.size() -> NumEltsPerBlock, as it's used more consistently.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D86225/new/

https://reviews.llvm.org/D86225