[PATCH] D144458: [RISCV] Pre-commit test case for ordered reduction, NFC

Philip Reames via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Wed Feb 22 15:12:25 PST 2023


reames added inline comments.


================
Comment at: llvm/test/Transforms/LoopVectorize/RISCV/ordered-reduction.ll:3
+; RUN: opt -passes=loop-vectorize -hints-allow-reordering=false -S < %s | FileCheck %s --check-prefix=CHECK-NOT-VECTORIZED
+; RUN: opt -passes=loop-vectorize -S < %s | FileCheck %s --check-prefix=CHECK-UNORDERED
+; RUN: opt -passes=loop-vectorize -force-ordered-reductions=false -hints-allow-reordering=false -S < %s | FileCheck %s --check-prefix=CHECK-NOT-VECTORIZED
----------------
reames wrote:
> Please add a check line for the configuration: -force-ordered-reductions=true -hints-allow-reordering=false.
> 
> This allows you to test the codegen for ordered reductions in this change and next becomes only changing which check prefixes one line uses.  
Also, this has nothing to do with your change, but the fact that -hints-allow-reordering defaults to true is deeply and unpleasantly surprising.  That is *not* the semantics I'd have expected from the loop directives.  


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144458/new/

https://reviews.llvm.org/D144458



More information about the llvm-commits mailing list