<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 - [NPM] LoopIdiom missed opportunity"
href="https://bugs.llvm.org/show_bug.cgi?id=48715">48715</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>[NPM] LoopIdiom missed opportunity
</td>
</tr>
<tr>
<th>Product</th>
<td>new-bugs
</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>new bugs
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>sjoerd.meijer@arm.com
</td>
</tr>
<tr>
<th>CC</th>
<td>htmldeveloper@gmail.com, llvm-bugs@lists.llvm.org
</td>
</tr></table>
<p>
<div>
<pre>Created <span class=""><a href="attachment.cgi?id=24368" name="attach_24368" title="Reproducer">attachment 24368</a> <a href="attachment.cgi?id=24368&action=edit" title="Reproducer">[details]</a></span>
Reproducer
Compile attached reproducer loop idiom with:
clang --target=arm-arm-none-eabi -mcpu=cortex-m33 -Ofast loopidiom.c -S
-Rpass=loop-idiom -fno-experimental-new-pass-manager
shows:
loopidiom.c:156:15: remark: Transformed loop-strided store into a call to
llvm.memset.p0i8.i32() function [-Rpass=loop-idiom]
*px++ = (q31_t)(sum1 >> 31);
^
whereas with the NPM and -fexperimental-new-pass-manager this memset is not
recognised.
I appreciate this is not a very detailed bug report, and I could have reduced
and analysed this further. But I am a bit in a hurry with this as I am looking
at a lot of changes and regressions (in different benchmarks). I also don't
know yet if this is the (only) root cause, but this kernel regresses with 20%
with the NPM and it's the first obvious change I see.</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>