<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html><head>
<meta name="generator" content="HTML Tidy, see www.w3.org">
<meta content="text/html; charset=ISO-8859-1" http-equiv="content-type"><title>Defmod</title></head><body>
PETSc's sparse MatMult performance for a matrix arising from unstructured FE linear-elasticity problem:<br>
<br>
<ul>
<li>For a 2D, 250K DOF problem on a:</li>
</ul>
<table style="text-align: left; width: 720px; height: 184px; margin-left: auto; margin-right: auto;" border="1" cellpadding="2" cellspacing="2">
<tbody>
<tr>
<td colspan="6" rowspan="1" style="vertical-align: top; text-align: center;">Single 2.6
GHz Intel Core i5-750 processor with
DDR3-1333 RAM running 32-bit Debian Linux with turbo boost disabled:</td>
</tr>
<tr>
<td colspan="1" rowspan="2" style="vertical-align: top;">Compiler
(Flags)<br>
</td>
<td colspan="2" rowspan="1" style="vertical-align: top; text-align: center;">Single core<br>
</td>
<td colspan="2" rowspan="1" style="vertical-align: top; text-align: center;">Quad Core<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">GFlop/sec<br>
</td>
<td style="vertical-align: top;">% of Peak<br>
</td>
<td style="vertical-align: top;">GFlop/sec</td>
<td style="vertical-align: top;">% of Peak<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">GCC 4.7 (-O3)<br>
</td>
<td style="vertical-align: top;">1.089</td>
<td style="vertical-align: top;">10.235<br>
</td>
<td style="vertical-align: top;">1.590</td>
<td style="vertical-align: top;"><span style="color: red;">3.736</span><br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">Intel XE 2013 (-O3)<br>
</td>
<td style="vertical-align: top;">1.155<br>
</td>
<td style="vertical-align: top;"><span style="color: red;">10.855</span><br>
</td>
<td style="vertical-align: top;">1.573<br>
</td>
<td style="vertical-align: top;">3.696<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">Open64 5.0 (-O2)<br>
</td>
<td style="vertical-align: top;">0.894<br>
</td>
<td style="vertical-align: top;">8.402<br>
</td>
<td style="vertical-align: top;">1.545<br>
</td>
<td style="vertical-align: top;">3.632<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">Solaris Studio 12.3 (-O3)<br>
</td>
<td style="vertical-align: top;">0.841<br>
</td>
<td style="vertical-align: top;">7.904<br>
</td>
<td style="vertical-align: top;">1.435<br>
</td>
<td style="vertical-align: top;">3.372<br>
</td>
</tr>
<tr>
<td colspan="6" rowspan="1" style="vertical-align: top; text-align: center;">Single 2.6 GHz AMD Istanbul processor
with DDR2-800 RAM running the 64-bit Cray Linux Environment:</td>
</tr>
<tr>
<td colspan="1" rowspan="2" style="vertical-align: top;">Compiler
(Flags)<br>
</td>
<td colspan="2" rowspan="1" style="vertical-align: top; text-align: center;">Single core<br>
</td>
<td colspan="2" rowspan="1" style="vertical-align: top; text-align: center;">Hex Core<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">GFlop/sec<br>
</td>
<td style="vertical-align: top;">% of Peak<br>
</td>
<td style="vertical-align: top;">GFlop/sec</td>
<td style="vertical-align: top;">% of Peak<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">Cray (vendor optimized)<br>
</td>
<td style="vertical-align: top;">0.785<br>
</td>
<td style="vertical-align: top;">7.548<br>
</td>
<td style="vertical-align: top;">1.281</td>
<td style="vertical-align: top;">2.053<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">GCC (vendor optimized)<br>
</td>
<td style="vertical-align: top;">0.769<br>
</td>
<td style="vertical-align: top;">7.394<br>
</td>
<td style="vertical-align: top;">1.283<br>
</td>
<td style="vertical-align: top;">2.056<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">Intel (vendor optimized)<br>
</td>
<td style="vertical-align: top;">0.579<br>
</td>
<td style="vertical-align: top;">5.567<br>
</td>
<td style="vertical-align: top;">1.134<br>
</td>
<td style="vertical-align: top;">1.817<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">PGI (vendor optimized)<br>
</td>
<td style="vertical-align: top;">0.600<br>
</td>
<td style="vertical-align: top;">5.769<br>
</td>
<td style="vertical-align: top;">1.109<br>
</td>
<td style="vertical-align: top;">1.777<br>
</td>
</tr>
<tr align="center">
<td colspan="6" rowspan="1" style="vertical-align: top;">Single
3.33 GHz Intel Westmere processor with
DDR3-1333 RAM running 64-bit CentOS Linux with HT disabled:<br>
</td>
</tr>
<tr>
<td colspan="1" rowspan="2" style="vertical-align: top;">Compiler
(Flags)<br>
</td>
<td colspan="2" rowspan="1" style="vertical-align: top; text-align: center;">Single core<br>
</td>
<td colspan="2" rowspan="1" style="vertical-align: top; text-align: center;">Hex Core<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">GFlop/sec<br>
</td>
<td style="vertical-align: top;">% of Peak<br>
</td>
<td style="vertical-align: top;">GFlop/sec</td>
<td style="vertical-align: top;">% of Peak<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">Intel 11.1 Compilers (-O3)<br>
</td>
<td style="vertical-align: top;"><span style="color: red;">1.233</span><br>
</td>
<td style="vertical-align: top;">9.257<br>
</td>
<td style="vertical-align: top;"><span style="color: red;">2.110</span><br>
</td>
<td style="vertical-align: top;">2.640<br>
</td>
</tr>
</tbody>
</table>
<br>
<ul>
<li>For a 3D, 1.5M DOF problemon on a:</li>
</ul>
<table style="text-align: left; width: 720px; height: 184px; margin-left: auto; margin-right: auto;" border="1" cellpadding="2" cellspacing="2">
<tbody>
<tr>
<td colspan="3" rowspan="1" style="vertical-align: top; text-align: center;">Single 2.6
GHz Intel Core i5-750 processor with
DDR3-1333 RAM running 32-bit Debian Linux with turbo boost disabled:</td>
</tr>
<tr>
<td colspan="1" rowspan="2" style="vertical-align: top;">Compiler
(Flags)<br>
</td>
<td colspan="2" rowspan="1" style="vertical-align: top; text-align: center;">Quad Core<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">GFlop/sec</td>
<td style="vertical-align: top;">% of Peak<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">GCC 4.7 (-O3)<br>
</td>
<td style="vertical-align: top;">2.489<br>
</td>
<td style="vertical-align: top;">5.983<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">Intel XE 2013 (-O3)<br>
</td>
<td style="vertical-align: top;">2.600<br>
</td>
<td style="vertical-align: top;"><span style="color: red;">6.250</span><br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">Open64 5.0 (-O2)<br>
</td>
<td style="vertical-align: top;">2.447<br>
</td>
<td style="vertical-align: top;">5.882<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">Solaris Studio 12.3 (-O3)<br>
</td>
<td style="vertical-align: top;">2.328<br>
</td>
<td style="vertical-align: top;">5.596<br>
</td>
</tr>
<tr>
<td colspan="3" rowspan="1" style="vertical-align: top; text-align: center;">Single 2.6 GHz AMD Istanbul processor
with DDR2-800 RAM running the 64-bit Cray Linux Environment:</td>
</tr>
<tr>
<td colspan="1" rowspan="2" style="vertical-align: top;">Compiler
(Flags)<br>
</td>
<td colspan="2" rowspan="1" style="vertical-align: top; text-align: center;">Hex Core<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">GFlop/sec</td>
<td style="vertical-align: top;">% of Peak<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">Cray (vendor optimized)<br>
</td>
<td style="vertical-align: top;">1.423</td>
<td style="vertical-align: top;">2.280<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">GCC (vendor optimized)<br>
</td>
<td style="vertical-align: top;">1.420<br>
</td>
<td style="vertical-align: top;">2.276<br>
</td>
</tr>
<tr align="center">
<td colspan="3" rowspan="1" style="vertical-align: top;">Single
3.33 GHz Intel Westmere processor with
DDR3-1333 RAM running 64-bit CentOS Linux with HT disabled:<br>
</td>
</tr>
<tr>
<td colspan="1" rowspan="2" style="vertical-align: top;">Compiler
(Flags)<br>
</td>
<td colspan="2" rowspan="1" style="vertical-align: top; text-align: center;">Hex Core<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">GFlop/sec</td>
<td style="vertical-align: top;">% of Peak<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">Intel 11.1 Compilers (-O3)<br>
</td>
<td style="vertical-align: top;"><span style="color: red;">3.070</span><br>
</td>
<td style="vertical-align: top;">3.841<br>
</td>
</tr><tr align="center">
<td colspan="3" rowspan="1" style="vertical-align: top;">Single 2.4 GHz AMD Magny Cours processor with DDR3-1333 running 64-bit Linux<br>
</td>
</tr>
<tr>
<td colspan="1" rowspan="2" style="vertical-align: top;">Compiler (Flags)<br>
</td>
<td colspan="2" rowspan="1" style="vertical-align: top; text-align: center;">Oct Core<br>
</td>
</tr>
<tr>
<td style="vertical-align: top;">GFlop/sec</td>
<td style="vertical-align: top;">% of Peak</td>
</tr>
<tr>
<td style="vertical-align: top;">PGI (none)<br>
</td>
<td style="vertical-align: top;">1.839<br>
</td>
<td style="vertical-align: top;">2.395<br>
</td>
</tr>
</tbody>
</table>
<br>
<span style="font-family: monospace;"></span>
</body></html>