<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; color: rgb(0, 0, 0); font-size: 14px; font-family: Calibri, sans-serif; ">
<div>
<div>
<div>In that case, I recommend that you those three GPU nodes plus any number of CPU nodes that make sense (with one MPI process per core). ParaView will distribute the rendering over all the nodes, GPU and CPU, even though you will only see output on those
6 processes. If your CPU nodes are set up similarly to your GPU nodes, that is with X hosts and OpenGL drivers, you just run all the processes the same. If your CPUs don't have X host, you can still perform rendering with Mesa 3D. In that case you'll need
to use different pvserver executables for each. ParaView won't actually understand the heterogeneity of your system, which will lead to some load imbalance, but not so much that I would bother to worry about it.</div>
<div><br>
</div>
<div>Does that answer your question?</div>
<div><br>
</div>
<div>-Ken</div>
<div><font size="2"><font face="Consolas,Courier New,Courier"><span style="font-size:10pt"><br>
**** Kenneth Moreland<br>
*** Sandia National Laboratories<br>
*********** <br>
*** *** *** email: <font color="#0000FF"><u><a href="kmorel@sandia.gov">kmorel@sandia.gov</a><br>
</u></font>** *** ** phone: (505) 844-8919<br>
*** web: <font color="#0000FF"><u><a href="http://www.cs.unm.edu/~kmorel">http://www.cs.unm.edu/~kmorel</a></u></font><br>
</span></font></font></div>
</div>
</div>
<div><br>
</div>
<span id="OLK_SRC_BODY_SECTION">
<div style="font-family:Calibri; font-size:11pt; text-align:left; color:black; BORDER-BOTTOM: medium none; BORDER-LEFT: medium none; PADDING-BOTTOM: 0in; PADDING-LEFT: 0in; PADDING-RIGHT: 0in; BORDER-TOP: #b5c4df 1pt solid; BORDER-RIGHT: medium none; PADDING-TOP: 3pt">
<span style="font-weight:bold">From: </span>Onat Yiğit Mercan <<a href="mailto:onat.mercan@be.itu.edu.tr">onat.mercan@be.itu.edu.tr</a>><br>
<span style="font-weight:bold">Date: </span>Sat, 27 Aug 2011 02:37:39 +0300<br>
<span style="font-weight:bold">To: </span>Kenneth Moreland <<a href="mailto:kmorel@sandia.gov">kmorel@sandia.gov</a>>, "<a href="mailto:paraview@paraview.org">paraview@paraview.org</a>" <<a href="mailto:paraview@paraview.org">paraview@paraview.org</a>><br>
<span style="font-weight:bold">Subject: </span>Re: [Paraview] Performance Test: Gpu vs. Cpu Rendering<br>
</div>
<div><br>
</div>
Sorry, I should be more clear. The system has 3 servers which are HP xw8600 workstations (<a href="http://h20331.www2.hp.com/Hpsub/downloads/xw8600_xw6600_FAQ_hi_res.pdf" target="_blank">http://h20331.www2.hp.com/Hpsub/downloads/xw8600_xw6600_FAQ_hi_res.pdf</a>).
Each of them has 1 gpu (nvidia quadro fx 5600 with 2 dvi connectors). There is only one cluster. Sorry about the confusion.
<div><br>
</div>
<div>I run 2 process for each gpu to achieve tiled display.<br>
<br>
<div class="gmail_quote">2011/8/26 Moreland, Kenneth <span dir="ltr"><<a href="mailto:kmorel@sandia.gov" target="_blank">kmorel@sandia.gov</a>></span><br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Sorry, but I'm confused by the question.<br>
<br>
You say you have a display comprising 6 screens. Your rendering cluster<br>
has 3 nodes each with 1 GPU. How do you get a signal to all 6 screens<br>
with only 3 GPU's? Does each GPU have a dual output that can drive two<br>
screens? If so, are the two screens treated as one so that one window<br>
from one MPI process covers both, or do you have to run two MPI processes<br>
per GPU node to drive both screens?<br>
<br>
You also say that you have nodes with only CPU's in them. How many do you<br>
have? And how are they connected to the GPU nodes? Are they two totally<br>
separate clusters or all they all part of the same cluster and the GPU<br>
nodes are just three special nodes in a larger cluster?<br>
<br>
-Ken<br>
<br>
**** Kenneth Moreland<br>
*** Sandia National Laboratories<br>
***********<br>
*** *** *** email: <a href="mailto:kmorel@sandia.gov" target="_blank">kmorel@sandia.gov</a><br>
** *** ** phone: <a href="tel:%28505%29%20844-8919" value="+15058448919" target="_blank">
(505) 844-8919</a><br>
*** web: <a href="http://www.cs.unm.edu/~kmorel" target="_blank">http://www.cs.unm.edu/~kmorel</a><br>
<br>
<br>
<br>
<br>
On 8/25/11 4:58 PM, "<a href="mailto:onat.mercan@be.itu.edu.tr" target="_blank">onat.mercan@be.itu.edu.tr</a>"<br>
<div>
<div></div>
<div><<a href="mailto:onat.mercan@be.itu.edu.tr" target="_blank">onat.mercan@be.itu.edu.tr</a>> wrote:<br>
<br>
>Hello,<br>
><br>
>I am working on blood flow visualization of cartoid artery. We have a<br>
>display wall which has 6 screens (2rows*3columns) and our system has 3<br>
>rendering servers (each of them has 1 gpu). As far as i know using tiled<br>
>screen mode let Paraview render on gpus and we can see the animation on<br>
>the screens and on cpu rendering mode, the only animation happens on<br>
>remote computer which is connected to servers.Is there any chance to see<br>
>animation on the screens while rendering on cpus because at the<br>
>performance comparison step, i see that cpu rendering has 25% rating than<br>
>gpu rendering (performance test has been done for 3gpus vs 3cpus/cores).<br>
>If you think that my problem is irrevelant to compare cpu and gpu<br>
>performances, what do you suggest?<br>
><br>
><br>
>--<br>
>This message has been scanned for viruses and<br>
>dangerous content by MailScanner, and is<br>
>believed to be clean.<br>
><br>
>_______________________________________________<br>
>Powered by <a href="http://www.kitware.com" target="_blank">www.kitware.com</a><br>
><br>
>Visit other Kitware open-source projects at<br>
><a href="http://www.kitware.com/opensource/opensource.html" target="_blank">http://www.kitware.com/opensource/opensource.html</a><br>
><br>
>Please keep messages on-topic and check the ParaView Wiki at:<br>
><a href="http://paraview.org/Wiki/ParaView" target="_blank">http://paraview.org/Wiki/ParaView</a><br>
><br>
>Follow this link to subscribe/unsubscribe:<br>
><a href="http://www.paraview.org/mailman/listinfo/paraview" target="_blank">http://www.paraview.org/mailman/listinfo/paraview</a><br>
><br>
<br>
<br>
--<br>
This message has been scanned for viruses and<br>
dangerous content by MailScanner, and is<br>
believed to be clean.</div>
</div>
</blockquote>
</div>
<br>
</div>
</span>
</body>
</html>