Xorg using a full core

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi, I have been testing out a Cuda wrapper for WINE to make the Folding@Home GPU Windows client run on Linux.

You can get up to speed here

[url=“http://foldingforum.org/viewtopic.php?f=43&t=3744&st=0&sk=t&sd=a&sid=288994a2006f70329c25d7fe38e967e1”]http://foldingforum.org/viewtopic.php?f=43...c25d7fe38e967e1[/url]

We have the client running but it seem that not only does the FAH process use up one full core but so does Xorg. This is a dual core system. When Xorg is requesting this much CPU time it basically locks up the computer. The client still runs until stopped but the computer is mostly useless during that time.

Why is Xorg using a full core during this? Are the Cuda Linux libraries currently out of sync with the Windows version in any way? This is similar to how this client acted in Windows before a few driver updates were released. Can we expect any similar optimizations for Linux any time soon?

Thank you.

2

CUDA releases are built from the same source for all operating systems, however DirectX fixes are obviously not going to be part of Linux CUDA releases.

This problem could be a wine bug, an X bug, or an nvidia X driver bug. It might be best if you followed the display driver README’s bug reporting instructions.

3

Thank you, that helps a lot. I was also planning on addressing this with the WINE guys. I will look into putting together a bug report once we have more information.