<?xml version="1.0"?>
<?xml-stylesheet type="text/xsl" href="/rss.xsl.xml"?>
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/">
<channel>
    <title>Changes in catchswitch.ll</title>
    <description></description>
    <language>en</language>
    <copyright>Copyright 2015</copyright>
    <generator>Java</generator><item>
        <title>bccc9aa8 - Don&apos;t vectorize PHIs in catchswitch blocks</title>
        <link>http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/test/Transforms/SLPVectorizer/X86/catchswitch.ll#bccc9aa8</link>
        <description>Don&apos;t vectorize PHIs in catchswitch blocksWe currently assert in vectorizeTree(TreeEntry*) when processing a PHIbundle in a block containing a catchswitch.  We attempt to set theIRBuilder insertion point following the catchswitch, which is invalid.This is done so that ShuffleBuilder.finalize() knows where to inserta shuffle if one is needed.To avoid this occurring, watch out for catchswitch blocks duringbuildTree_rec() processing, and avoid adding PHIs in such blocks tothe vectorizable tree.  It is unlikely that constraining vectorizationover an exception path will cause a noticeable performance loss, sothis seems preferable to trying to anticipate when a shuffle will andwill not be required.

            List of files:
            /llvm-project-15.0.7/llvm/test/Transforms/SLPVectorizer/X86/catchswitch.ll</description>
        <pubDate>Mon, 18 Jul 2022 20:55:27 +0000</pubDate>
        <dc:creator>William Schmidt &lt;william.schmidt@intel.com&gt;</dc:creator>
    </item>
</channel>
</rss>
