<div dir="ltr">To give an idea of the changes, I uploaded: <a href="https://reviews.llvm.org/D73592">https://reviews.llvm.org/D73592</a>.<br><div><br></div><div>Thanks,</div><div>Alina</div><div><br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Jan 28, 2020 at 2:44 PM Alina Sbirlea <<a href="mailto:alina.sbirlea@gmail.com">alina.sbirlea@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">Hi,<div><br></div><div>Recently the LoopUnrollAndJamPass was made a function pass, and here was the motivation for that: <a href="https://groups.google.com/forum/#!topic/llvm-dev/LF4rUjkVI2g" target="_blank">https://groups.google.com/forum/#!topic/llvm-dev/LF4rUjkVI2g</a>.</div><div>I don't see an obvious reason to keep IRCE as a LoopPass either, so I thought folks familiar with the pass could chime in if this seems like the right change.</div><div>An additional motivation is the need to either preserve BranchProbabilityInfo or have a new instance computed for each Loop (<a href="https://reviews.llvm.org/D72891" target="_blank">D72891</a>). Making IRCE a FunctionPass will address this issue.</div><div><br></div><div>Thanks,</div><div>Alina</div>
</div>
</blockquote></div>