<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 - [meta][x86] Scheduler Driven Optimizations"
href="https://bugs.llvm.org/show_bug.cgi?id=35784">35784</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>[meta][x86] Scheduler Driven Optimizations
</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>Backend: X86
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>llvm-dev@redking.me.uk
</td>
</tr>
<tr>
<th>CC</th>
<td>andrea.dibiagio@gmail.com, andrew.v.tischenko@gmail.com, craig.topper@gmail.com, gadi.haber@intel.com, llvm-bugs@lists.llvm.org, spatel+llvm@rotateright.com, zvi.rackover@intel.com
</td>
</tr>
<tr>
<th>Blocks</th>
<td>32325
</td>
</tr></table>
<p>
<div>
<pre>Split off from <a class="bz_bug_link
bz_status_NEW "
title="NEW - Combining unary integer shuffles to PSHUFB"
href="show_bug.cgi?id=26183">Bug #26183</a>.
With the scheduler models maturing we should be investigating ways to better
select between MC optimizations based on scheduler characteristics, register
pressure etc.
<a class="bz_bug_link
bz_status_NEW "
title="NEW - Combining unary integer shuffles to PSHUFB"
href="show_bug.cgi?id=26183">Bug #26183</a> mentioned several X86 slow/fast features that could be replaced with
scheduler costings:
FeatureSlowBTMem
FeatureSlowSHLD
FeatureSlowUAMem16
FeatureSlowUAMem32
FeatureSlowDivide32
FeatureSlowDivide64
Some of these affect (older) CPUs that currently don't have a accurate
scheduler models - taking numbers from Agner is probably enough to be useful?
Additionally, the use of when to combine multiple shuffles to a single variable
shuffle mask keeps coming up. As well as the actual instruction cost, the cost
of loading the mask needs to be accurately modelled: register pressure? can it
be hoisted? etc.</pre>
</div>
</p>
<div id="referenced">
<hr style="border: 1px dashed #969696">
<b>Referenced Bugs:</b>
<ul>
<li>
[<a class="bz_bug_link
bz_status_NEW "
title="NEW - [META][X86] Improve implementation and use of X86 scheduler models"
href="https://bugs.llvm.org/show_bug.cgi?id=32325">Bug 32325</a>] [META][X86] Improve implementation and use of X86 scheduler models
</li>
</ul>
</div>
<br>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are on the CC list for the bug.</li>
</ul>
</body>
</html>