[PATCH] D108115: [llvm][sve] Lowering for VLS masked truncating stores

[David Truby via Phabricator via llvm-commits]

DavidTruby added inline comments.


================
Comment at: llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp:9822
+  // truncating store.  We can do this even if this is already a truncstore.
+  if ((Value.getOpcode() == ISD::FP_ROUND ||
+       Value.getOpcode() == ISD::TRUNCATE) &&
----------------
dmgreen wrote:
> I'm surprised to see FP_ROUND here. I guess it gets handled in the same way as a normal non-masked trunc store?
I just took this from the existing non-masked truncating store code. I think it gets handled by canCombineTruncStore checking if the truncating store has been marked legal or custom, which it won't have been if the architecture doesn't support floating point truncating stores.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D108115/new/

https://reviews.llvm.org/D108115