[PATCH] D143283: [AArch64][SVE]: custom lower AVGFloor/AVGCeil.

[Dave Green via Phabricator via llvm-commits]

dmgreen added a comment.

Thanks. here are some alive proofs for the transform in https://alive2.llvm.org/ce/z/N6hwQY and https://alive2.llvm.org/ce/z/u_GjYJ.

Can you extend the testing to include both ashr and lshr versions? They should both be useful if we are custom legalizing the nodes. Otherwise I think this looks good.



================
Comment at: llvm/test/CodeGen/AArch64/sve-hadd.ll:25
   %m = add nsw <vscale x 2 x i128> %s0s, %s1s
-  %s = lshr <vscale x 2 x i128> %m, shufflevector (<vscale x 2 x i128> insertelement (<vscale x 2 x i128> poison, i128 1, i32 0), <vscale x 2 x i128> poison, <vscale x 2 x i32> zeroinitializer)
+  %s = ashr <vscale x 2 x i128> %m, shufflevector (<vscale x 2 x i128> insertelement (<vscale x 2 x i128> poison, i128 1, i32 0), <vscale x 2 x i128> poison, <vscale x 2 x i32> zeroinitializer)
   %s2 = trunc <vscale x 2 x i128> %s to <vscale x 2 x i64>
----------------
Can you copy these tests so there are versions with both lshr and ashr.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D143283/new/

https://reviews.llvm.org/D143283