<div dir="ltr"><div class="gmail_extra">And what is the performance improvement?<br><br><div class="gmail_quote">On Fri, Apr 22, 2016 at 11:37 AM, Sebastian Pop <span dir="ltr"><<a href="mailto:sebpop@gmail.com" target="_blank">sebpop@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">sebpop removed rL LLVM as the repository for this revision.<br>
</span>sebpop updated this revision to Diff 54684.<br>
sebpop added a comment.<br>
<br>
The updated patch from Aditya has some cleanups, compile time improvements, and uses less memory.<br>
<br>
We measured the total number of instructions executed when "clang -cc1" is compiling all the preprocessed files of the llvm test-suite through valgrind:<br>
<br>
- with patch: 926457230716<br>
- without: 923166450214<br>
<br>
Overall compilation overhead of the new pass is 0.35%.<br>
<div class="HOEnZb"><div class="h5"><br>
<br>
<a href="http://reviews.llvm.org/D19338" rel="noreferrer" target="_blank">http://reviews.llvm.org/D19338</a><br>
<br>
Files:<br>
  llvm/include/llvm/InitializePasses.h<br>
  llvm/include/llvm/LinkAllPasses.h<br>
  llvm/include/llvm/Transforms/Scalar.h<br>
  llvm/include/llvm/Transforms/Scalar/GVN.h<br>
  llvm/lib/Passes/PassRegistry.def<br>
  llvm/lib/Transforms/IPO/PassManagerBuilder.cpp<br>
  llvm/lib/Transforms/Scalar/CMakeLists.txt<br>
  llvm/lib/Transforms/Scalar/GVNHoist.cpp<br>
  llvm/lib/Transforms/Scalar/Scalar.cpp<br>
  llvm/test/Transforms/GVN/hoist.ll<br>
<br>
</div></div></blockquote></div><br></div></div>