[info-performer] Running Performer on a cluster node

From: Jean-Francois Dupuis (jfdupuis++at++gel.ulaval.ca)
Date: 02/18/2005 09:12:44

• Next message: faculaganymede: "[info-performer] glDrawPixels n pfuSaveImage Conflicts"
• Previous message: Moskalski Timothy J CRBE: "[info-performer] wavefront obj loader"

Hi,

I would like to run an OpenGL Performer 3.2.0 application on a cluster
node, which have no video card and no input device. The images generated
by Performer are used by an image processing routine. Thus, I tried to
run my application using the X Virtual Frame Buffer (Xvfb) and reading
image with glReadPixels in a Draw Callback.

Currently, my application run well on a linux workstation running the
default X server coming with Fedora Core 1. However, when I run the same
application using Xvfb, the rendering is incomplete. The bottom part of
the image is good, but the upper part look good but is blurred.

This also happend if I run the application over ssh on a workstation on
which Performer is installed and displaying the output on a workstation
on which Performer is not installed. However, this do not happend if I
run the application with two workstation on which Performer is
installed.

If this is not the righ way I should process to run Performer on a
cluster node, can someone indicate me the righ way to go ?

Thanks,

Jean-Francois Dupuis

• Next message: faculaganymede: "[info-performer] glDrawPixels n pfuSaveImage Conflicts"
• Previous message: Moskalski Timothy J CRBE: "[info-performer] wavefront obj loader"

This archive was generated by hypermail 2b29 : Fri Feb 18 2005 - 09:13:01 PST