<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 - [X86] Could commute some VEX encoded instructions to enable 2 byte VEX prefix"
href="https://bugs.llvm.org/show_bug.cgi?id=36706">36706</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>[X86] Could commute some VEX encoded instructions to enable 2 byte VEX prefix
</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>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>craig.topper@gmail.com
</td>
</tr>
<tr>
<th>CC</th>
<td>llvm-bugs@lists.llvm.org
</td>
</tr></table>
<p>
<div>
<pre>The VEX prefix comes in two forms, 2 byte and 3 byte. The 2 byte version is
missing some fields from the 3 byte version so can't encode all instructions.
The missing fields are VEX.W, VEX.mmmmm, VEX.X, and VEX.B. VEX.X is only used
by memory instructions. VEX.W isn't used on most instructions.
For a commutable instructions like the register only form of VPADDD, if one
input is xmm8-15 and the other is xmm0-7, we could take advantage of commuting
to always put the xmm8-15 in the operand that uses VEX.R to avoid needing
VEX.B.
This would only work for instructions that use VEX.mmmmm = 0x1, but that's the
more common encoding for that field.</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>