[PATCH] D135808: [LoopInterchange] Correcting the profitability checking for vectorization

[Michael Kruse via Phabricator via llvm-commits]

Meinersbur added a comment.

testcase <https://reviews.llvm.org/file/data/djr5c2jm67s4ejikk6a7/PHID-FILE-wmeppydlvqc2ag2awnkj/meme-test_case>



================
Comment at: llvm/lib/Transforms/Scalar/LoopInterchange.cpp:1175-1176
                                     InnerLoop->getHeader())
            << "Interchanging loops is too costly and it does not improve "
               "parallelism.";
   });
----------------
Consider updating this message. Suggestion: "Interchanging loops not considered to improve cache locality nor vectorization."


================
Comment at: llvm/lib/Transforms/Scalar/LoopInterchange.cpp:1166
+  // analyze the loopnest (e.g., due to delinearization issues).
+  if (CostMap.find(InnerLoop) == CostMap.end() || CostMap.find(OuterLoop) != CostMap.end() ||
+     (CC && CC->getLoopCost(*InnerLoop) == CC->getLoopCost(*OuterLoop))) {
----------------
ram-NK wrote:
> congzhe wrote:
> > 
> Corrected the condition. any of the loop is failed to determine the loop nest and equal locality then only checks the profit of vectorization.
This is exactly the else branch of the cache analysis logic. It does not consider the fallback `Cost < -LoopInterchangeCostThreshold` legacy cost model. Please avoid the code duplication and needing to loop up the `CostMap` again.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D135808/new/

https://reviews.llvm.org/D135808