Hello everyone,
Just a technical note.
of course, everyone has the right to post in the forum he or she wants, but:

Should this post not be in the suggestions and feedback forum?

One more thing:
I wish, the community adivsers had the technical possiblity to transfer posts to the forum in which they sould have been posted.

Of course, the member who postet the post in the wrong forum should be informed about the transfer of his or her post to a different forum. There could be with an automatical note in the forum, in which the post had been originally posted, This note could be for example: Dear Member, the post you posted here three hours ago, has been transferred to the xy forum. This note will dissappear in x day.

The transfer notes should ineed dissappear after a week, because otherwise they would disturb.

All the best to erveryone
Martin S

Sekerob : The setup presented by Saul Luizaga is necessary when using several cards, and that, just like setting up a server farm, is for the tech savvy. Their GPU client is not perfect but stable.

For a regular machine, just download the client, register with F@H and it'll start crunching.
http://folding.stanford.edu/English/DownloadWinOther
The work they're doing is about protein folding (different, but in the same field as HPF2). WCG & F@H happily coexist on the same machine.

The only drawbacks are :
- higher fan noise
- need to clean the card's fan a bit more often
- No side effect for most applications, but marked slowdown for a few such as Google Sketchup (at least on my machine), but there is a "Pause Work" option on the tray icon.

Checking GPU temperature can be done with a simple utility such as :
http://www.techpowerup.com/downloads/1898/TechPowerUp_GPU-Z_v0.4.8.html
Temps around 80 deg. Centigrade are normal for the GPU itself, check at least once every 3 months or if noise from the GPU fan increases (needs a bit of cleanup).

The CPU part of the application uses below 3%, so the whole CPUs remain usable for WCG projects.

My main machine is a Win XP Q9550 + GTX275 and it's been crunching for both projects for about 2 years without incident.

djibeXX wrote:
"Thanks a lot. Nvidia and ATI will soon be leaders in supercomputers ... or I hope so."

At least, on a symbolic level, it already happened, as the fastest supercomputer in the latest "TOP500" ranking is the Chinese Tianhe-1A, made using NVidia GPUs :
http://www.top500.org/lists/2010/11/press-release

Ingleside wrote:

Saul Luizaga wrote it used 5% of a quad core with ATI, and my experience is around 3% with NVidia. If you have a a slower machine, the relative percentage will be higher, else there must be something wrong with your specific machine and you might get better advice on their forum at http://foldingforum.org/

The point was that it would be better if WCG had some GPU projects to keep our machines warm, but since this is quite unlikely in the foreseeable future, there are other worthwhile projects out there that behave well with WCG.

OTOH, since F@H computation is so much faster & power efficient in GPU mode, it would be a better use of computing power if people currently crunching for F@H with a CPU only machine could crunch for WCG instead.

Such cooperation between projects could bring huge benefits to both, as running their CPU client is highly inefficient at what they do (10% of their users are GPU users, yet they contribute 90% of the processing power)

Here is the average power (raw data : http://fah-web.stanford.edu/cgi-bin/main.py?qtype=osstats) :
NVidia : 335 Mflop
ATI : 149 Mflop
PS3 : 59 Mflop
Windows CPU : 1.05 Mflop
Linux CPU : 2.7 Mflop

Total contribution is 812 Tflops for 470000 active CPUs versus 9447 TFlops for only 55000 active GPUs
On average, a single GPU contributes like around 100 CPUs

To be fair, the very high numbers are a result of both GPU power, highly optimized algorithm and longer average daily run time for those machines.

An "exchange" where WCG users with an idle GPU would run the F@H GPU client while F@H CPU only users would run WCG where their machine would have a more valuable contribution would bring huge benefits to both projects, but in the current competition for users, this would require either cooperation from the projects maintainers or a group of users from both projects willing to cross the border and maximize their global contribution.

Here is very interesting "client" for Folding@Home. Easier for beginners on this project and with a lot of options. Check it out ;)

While it would make sense to run A and B DDDT2 types on GPU's and C types on CPU's, I doubt that WCG and F@H would co-operate on this and infrastructure is not in place to do all of this at WCG.

OT, Why people would want to move their ATI's to count pi and search for Spock is one for the psychologists.