[PATCH] D125856: [RISCV] Add cost model for SK_Reverse

[LiqinWeng via Phabricator via llvm-commits]

Miss_Grape added inline comments.


================
Comment at: llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp:288
+  // The code-generator is currently not able to handle scalable vectors
+  // of <vscale x 1 x eltty> yet, so return an invalid cost to avoid selecting
+  // it. This change will be removed when code-generation for these types is
----------------
craig.topper wrote:
> There are no problems with <vscale x 1 x eltty> in the code generator that I'm aware of. Please explain.
Fixed


================
Comment at: llvm/test/Transforms/LoopVectorize/RISCV/riscv-vector-reverse.ll:3
+; REQUIRES: asserts
+; RUN: opt -loop-vectorize -dce -instcombine -mtriple riscv64-linux-gnu \
+; RUN:   -mattr=+v -debug-only=loop-vectorize \
----------------
craig.topper wrote:
> Shouldn't a reverse test have reverse intrinsics in it? Or did instcombine remove them?
pls review： https://reviews.llvm.org/D125866 


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D125856/new/

https://reviews.llvm.org/D125856