Hi, I am trying to understand how llvm's SLP vectorizer vectorizes reduction. I get the gist but don't see how phi nodes come into play here. Why does `matchAssociativeReduction` take an additional phi node as an argument? Thanks in advance, Tom -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20200826/069d73f9/attachment.html>