[PATCH] D135640: fix vectorization analysis msg for runtime checks

[Josh Milthorpe via Phabricator via llvm-commits]

milthorpe created this revision.
milthorpe added a reviewer: mkuper.
Herald added a subscriber: hiraditya.
Herald added a project: All.
milthorpe requested review of this revision.
Herald added a subscriber: pcwang-thead.
Herald added a project: LLVM.

Update analysis message to indicate that the loop was not vectorized
due to unprofitability from too many required runtime checks, not
because it is unsafe to do so.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D135640

Files:
  llvm/lib/Transforms/Vectorize/LoopVectorize.cpp


Index: llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
===================================================================
--- llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -10350,8 +10350,8 @@
         return OptimizationRemarkAnalysisAliasing(
                    DEBUG_TYPE, "CantReorderMemOps", L->getStartLoc(),
                    L->getHeader())
-               << "loop not vectorized: cannot prove it is safe to reorder "
-                  "memory operations";
+               << "loop not vectorized: it is not profitable to reorder memory "
+                  "operations with runtime checks";
       });
       LLVM_DEBUG(dbgs() << "LV: Too many memory checks needed.\n");
       Hints.emitRemarkWithHints();


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D135640.466691.patch
Type: text/x-patch
Size: 779 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20221011/c0035c06/attachment.bin>