<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<style type="text/css" style="display:none;"><!-- P {margin-top:0;margin-bottom:0;} --></style>
</head>
<body dir="ltr">
<div id="divtagdefaultwrapper" style="font-size:12pt;color:#000000;font-family:Calibri,Helvetica,sans-serif;" dir="ltr">
<p>Hi Michel,</p>
<p><br>
</p>
<p>I had looked at the issue and it seems variable indexing has always been somewhat broken in LLVM and this commit uncovered the bug.</p>
<p><br>
</p>
<p>Marek<br>
</p>
</div>
<hr style="display:inline-block;width:98%" tabindex="-1">
<div id="divRplyFwdMsg" dir="ltr"><font face="Calibri, sans-serif" style="font-size:11pt" color="#000000"><b>From:</b> Michel Dänzer <michel@daenzer.net><br>
<b>Sent:</b> Thursday, November 9, 2017 5:11:46 PM<br>
<b>To:</b> Olsak, Marek<br>
<b>Cc:</b> llvm-commits@lists.llvm.org<br>
<b>Subject:</b> Re: [llvm] r317753 - AMDGPU: Merge BUFFER_LOAD_DWORD_OFFSET into x2, x4</font>
<div> </div>
</div>
<div class="BodyFragment"><font size="2"><span style="font-size:10pt;">
<div class="PlainText"><br>
Hi Marek,<br>
<br>
<br>
On 09/11/17 02:52 AM, Marek Olsak via llvm-commits wrote:<br>
> Author: mareko<br>
> Date: Wed Nov 8 17:52:36 2017<br>
> New Revision: 317753<br>
> <br>
> URL: <a href="http://llvm.org/viewvc/llvm-project?rev=317753&view=rev">http://llvm.org/viewvc/llvm-project?rev=317753&view=rev</a><br>
> Log:<br>
> AMDGPU: Merge BUFFER_LOAD_DWORD_OFFSET into x2, x4<br>
<br>
This change broke a bunch of<br>
spec@glsl-1{1,2}0@execution@variable-indexing@vs-varying-array-mat4-index-*<br>
tests on my Tonga, e.g.<br>
<br>
PIGLIT TEST: 73 - vs-varying-array-mat4-index-col-wr<br>
Probe color at (10,10)<br>
Expected: 0.000000 1.000000 0.000000<br>
Observed: 1.000000 0.000000 0.000000<br>
Test failure on line 99<br>
Probe color at (25,10)<br>
Expected: 0.000000 1.000000 0.000000<br>
Observed: 1.000000 0.000000 0.000000<br>
Test failure on line 107<br>
Probe color at (40,10)<br>
Expected: 0.000000 1.000000 0.000000<br>
Observed: 1.000000 0.000000 0.000000<br>
Test failure on line 115<br>
[...]<br>
<br>
<br>
These tests consistently pass without this change and fail with it.<br>
<br>
<br>
I'm attaching the dump of an affected shader with and without this<br>
change. Let me know if you need more information.<br>
<br>
<br>
-- <br>
Earthling Michel Dänzer | <a href="http://www.amd.com">
http://www.amd.com</a><br>
Libre software enthusiast | Mesa and X developer<br>
</div>
</span></font></div>
</body>
</html>