en Copyright 2015 Java http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/mixed-precision-remarks.ll#c3a3d200 [LV] Add analysis remark for mixed precision conversionsFloating point conversions inside vectorized loops have performanceimplications but are very subtle. The user could specify a floatingpoint constant, or call a function without realizing that it willforce a change in the vector width. An example of this behaviour isseen in https://godbolt.org/z/M3nT6c . The vectorizer should indicatewhen this happens becuase it is most likely unintended behaviour.This patch adds a simple check for this behaviour by following floatingpoint stores in the original loop and checking if a floating pointconversion operation occurs.Reviewed By: fhahnDifferential Revision: https://reviews.llvm.org/D95539 List of files: /llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/mixed-precision-remarks.ll Wed, 27 Jan 2021 15:34:01 +0000 Joseph Huber <jhuber6@vols.utk.edu>