[Paraview] Questions about Paraview, renderserver and dataserver

[Moreland, Kenneth]

Recommendation number 1: 750,000 points is not really that large.  Rather than spend the time establishing a makeshift cluster, just load the data on your beefiest desktop and view it there.  I just created a 800,000 point data set on my laptop and it works reasonably well.

Recommendation number 2: If you insist on using client server, just use a single (multiprocess) pvserver instead of the dual pvrenderserver/pvdataserver.  See the discussion on the Wiki: http://www.paraview.org/Wiki/Setting_up_a_ParaView_Server#pvserver_vs._pvrenderserver_and_pvdataserver

-Ken

________________________________
From: paraview-bounces at paraview.org [mailto:paraview-bounces at paraview.org] On Behalf Of Cally K
Sent: Friday, April 18, 2008 11:47 PM
To: ParaView
Subject: [Paraview] Questions about Paraview, renderserver and dataserver

Hi, I would like to clarify some stuff about data server and render server


I plan to build paraview using this architecture :

Dataserver = 2 node cluster machine,
a)      Pentium(R) 4 CPU 3.00GHz - 2GB RAM
b)      Pentium 3 -256MB RAM
c) Hard disk


Renderserver = Quad Core Intel Xeon machine, 1GB RAM, 80GB hard disk

Paraview Client = Intel Pentium Dual Core, 160 GB Hard disk, 2GB RAM


a)Umm, is this ok .. or should I reassemble them... any idea..  Cause I have read that the render server should not consist of more nodes that the dataserver.. but doesn't the memory matter also?

b)I am using OpenMPI, do I need to install any batch scheduling program ..any suggestions..


c) I have a 750,000 points * 53 timesteps to render. And I found this fact on paraview wiki

1.         Sort first (collection): Small objects can be collected to the first node and rendered locally. In this mode, the polygons are collected only when the source generating them is modified. Although this collection might take time, it is done only once and rendering is fast.
2.         Sort last (distributed rendering): In this mode, first, each process renders their scene, then these scenes are composited using the depth buffer. This is done every render. Although this is slower for smaller datasets, it is the only scalable solution when the dataset become large
I don't really understand this thing, how does it apply to my data. I think it falls under the 2nd option because the dataset is large.. but I will be using only one quad core machine.. mentioned above


Any help.. Could someone point out to me ..? Thanks

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20080421/517c6605/attachment.htm>