<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html><head><title></title>
<META http-equiv=Content-Type content="text/html; charset=iso-8859-1">
<meta http-equiv="Content-Style-Type" content="text/css">
<style type="text/css"><!--
body {
margin: 5px 5px 5px 5px;
background-color: #ffffff;
}
/* ========== Text Styles ========== */
hr { color: #000000}
body, table /* Normal text */
{
font-size: 9pt;
font-family: 'Courier New';
font-style: normal;
font-weight: normal;
color: #000000;
text-decoration: none;
}
span.rvts1 /* Heading */
{
font-size: 10pt;
font-family: 'Arial';
font-weight: bold;
color: #0000ff;
}
span.rvts2 /* Subheading */
{
font-size: 10pt;
font-family: 'Arial';
font-weight: bold;
color: #000080;
}
span.rvts3 /* Keywords */
{
font-size: 10pt;
font-family: 'Arial';
font-style: italic;
color: #800000;
}
a.rvts4, span.rvts4 /* Jump 1 */
{
font-size: 10pt;
font-family: 'Arial';
color: #008000;
text-decoration: underline;
}
a.rvts5, span.rvts5 /* Jump 2 */
{
font-size: 10pt;
font-family: 'Arial';
color: #008000;
text-decoration: underline;
}
span.rvts6
{
font-weight: bold;
color: #ffffff;
background-color: #0000ff;
}
span.rvts7
{
}
span.rvts8
{
font-weight: bold;
color: #800000;
}
span.rvts9
{
font-weight: bold;
color: #800080;
}
a.rvts10, span.rvts10
{
color: #0000ff;
text-decoration: underline;
}
span.rvts11
{
background-color: #ffffff;
}
/* ========== Para Styles ========== */
p,ul,ol /* Paragraph Style */
{
text-align: left;
text-indent: 0px;
padding: 0px 0px 0px 0px;
margin: 0px 0px 0px 0px;
}
.rvps1 /* Centered */
{
text-align: center;
}
--></style>
</head>
<body>
<p>Hallo Jan,</p>
<p><br></p>
<p>Samstag, 19. April 2008, meintest Du:</p>
<p><br></p>
<div><a name="ctl00_MainContentPlaceholder_ctl01_ctl00_lblEntry"></a>
<table border=0 cellpadding=1 cellspacing=2>
<tr valign=top>
<td width=7 style="background-color: #0000ff;">
<p><span class=rvts6>></span></p>
</td>
<td width=1462 style="background-color: #ffffff;">
<p><br></p>
<p><span class=rvts8>KC> That's a pretty amusing blog, actually. See</span></p>
<p><a class=rvts10 href="http://terboven.spaces.live.com/blog/cns!EA3D3C756483FECB!255.entry">KC> http://terboven.spaces.live.com/blog/cns!EA3D3C756483FECB!255.entry</a><span class=rvts8> </span></p>
<p><span class=rvts8>KC> (wonderful permalinks, btw).</span></p>
<p><br></p>
<p><br></p>
<p><span class=rvts7>Read this arcticle to the end:</span></p>
<p><br></p>
<p><br></p>
<p><span class=rvts11>"Finally, lets take a quick look at some performance measurements. From the Intel MPI Benchmark Suite V3.1 I used the MPI-1 part to measure the latency and bandwidth. Please note that you have to divide the PingPong-latency by two in order to get the "real" MPI-latency. In doing so we see a MPI latency of aboud 2.9 us on our 4x DDR InfiniBand.</span><span class=rvts7> "</span></p>
<p><br></p>
<p><span class=rvts7>What? They divided the number of the benchmark by two? Correct me if i'm wrong but i was always assuming that the benchmark does this himself!</span></p>
<p><br></p>
<p><span class=rvts7>Can somebody confirm that? </span></p>
</td>
</tr>
</table>
</div>
<p><br></p>
<p>So i answer myself:</p>
<p><br></p>
<p><span class=rvts10>ftp://ftp.uybhm.itu.edu.tr/belgeler/sistem_kullanimi/IMB_ug-3.0.pdf</span></p>
<p><br></p>
<p>Page 10 Chapter 3.3.2.1 PingPong: "time = Δt/2 (in μsec) as indicated in Figure 1". So the tool already shows just the one-way-time. No division necessary.</p>
<p><br></p>
<p>Cheers,</p>
<p>Jan</p>
</body></html>