<html>
<head>
<base href="https://bugs.llvm.org/">
</head>
<body><table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>Bug ID</th>
<td><a class="bz_bug_link
bz_status_NEW "
title="NEW - MemorySSA could cache alias query results more aggressively"
href="https://bugs.llvm.org/show_bug.cgi?id=33126">33126</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>MemorySSA could cache alias query results more aggressively
</td>
</tr>
<tr>
<th>Product</th>
<td>libraries
</td>
</tr>
<tr>
<th>Version</th>
<td>trunk
</td>
</tr>
<tr>
<th>Hardware</th>
<td>PC
</td>
</tr>
<tr>
<th>OS</th>
<td>Windows NT
</td>
</tr>
<tr>
<th>Status</th>
<td>NEW
</td>
</tr>
<tr>
<th>Severity</th>
<td>enhancement
</td>
</tr>
<tr>
<th>Priority</th>
<td>P
</td>
</tr>
<tr>
<th>Component</th>
<td>Scalar Optimizations
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>davide@freebsd.org
</td>
</tr>
<tr>
<th>CC</th>
<td>dberlin@dberlin.org, george.burgess.iv@gmail.com, llvm-bugs@lists.llvm.org
</td>
</tr></table>
<p>
<div>
<pre>Dumping this one here, for whomever gets to it first (as I have a terrible
memory, and probably will forget in a day or two).
Dan run ./opt -memoryssa on clang.bc and statistics return:
1175012 memoryssa - Number of MemorySSA alias queries
360784 memoryssa - Number of MemorySSA duplicate pointer queries
Which is a pretty significant amount of duplicate queries.
In the conducted experiments, caching everything saves 20% of MemorySSA time,
but the amount of memory used could be a concern for large testcases (e.g.
LTO).
Probably using a smarter cache could lead to substantial benefits without
memory usage going crazy. It's unclear what's the best policy, maybe LRU is a
good starting point.</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are on the CC list for the bug.</li>
</ul>
</body>
</html>