<html>
<head>
<base href="http://llvm.org/bugs/" />
</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 --- - BlockFrequency underflow leads to slow code for 7zip"
href="http://llvm.org/bugs/show_bug.cgi?id=16402">16402</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>BlockFrequency underflow leads to slow code for 7zip
</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>All
</td>
</tr>
<tr>
<th>Status</th>
<td>NEW
</td>
</tr>
<tr>
<th>Severity</th>
<td>normal
</td>
</tr>
<tr>
<th>Priority</th>
<td>P
</td>
</tr>
<tr>
<th>Component</th>
<td>Global Analyses
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>benny.kra@gmail.com
</td>
</tr>
<tr>
<th>CC</th>
<td>atrick@apple.com, llvmbugs@cs.uiuc.edu, stoklund@2pi.dk
</td>
</tr>
<tr>
<th>Classification</th>
<td>Unclassified
</td>
</tr></table>
<p>
<div>
<pre>In LzmaEnc_CodeOneBlock in LzmaEnc.c from MultiSouce/Benchmarks/7zip the block
frequency analysis returns 0 frequencies for blocks in the inner loop. The
frequency calculation hits 0 at some point and never recovers. This made
r184105, the change that made spill weight depend on BlockFrequencyInfo,
regress 7zip by 10% due to excessive spilling. Simply raising
BlockFrequency::ENTRY_FREQ recovers the regression.
Can we do something about this? Increasing ENTRY_FREQ looks like a band aid to
me, we could hit this again any time.</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>