<html>
<head>
<base href="https://bugs.freedesktop.org/">
</head>
<body>
<p>
<div>
<b><a class="bz_bug_link
bz_status_NEW "
title="NEW - [BSW BXT GLK] dEQP-VK.subgroups.arithmetic.subgroup regressions"
href="https://bugs.freedesktop.org/show_bug.cgi?id=109328#c1">Comment # 1</a>
on <a class="bz_bug_link
bz_status_NEW "
title="NEW - [BSW BXT GLK] dEQP-VK.subgroups.arithmetic.subgroup regressions"
href="https://bugs.freedesktop.org/show_bug.cgi?id=109328">bug 109328</a>
from <span class="vcard"><a class="email" href="mailto:currojerez@riseup.net" title="Francisco Jerez <currojerez@riseup.net>"> <span class="fn">Francisco Jerez</span></a>
</span></b>
<pre>I see what's going on. Apparently the SHUFFLE virtual instruction used by
Vulkan 1.1 subgroups operations doesn't behave correctly when its sources are
aligned in a way that complies with the CHV/BXT restrictions for
double-precision datatypes, and it's going to have to be special-cased in the
regioning lowering pass.
Mark, is there some way I can force the CI to run on these "slow" platforms
when I foresee potential regressions, so this doesn't happen again in the
future?</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are the QA Contact for the bug.</li>
</ul>
</body>
</html>