Folding Forum

Hi,

To help the folding community, I would like to buy a new graphic card or a new processor that will be the most efficient as possible. I think the latest ones could be a good choice, but, what is the difference between an ATI Radeon HD 4870 or a NVIDIA Geforce GTX 280 ? Is the first one faster ? From which information can we say it's more or less fast to use the second one or the first one ?

I have no answer to this sort of question, and I think we can speak a lot about graphic cards and processors but it will not be significant. To have objective conversation, we will need a referential that allow us to compare every product on the market. Is it the pixels shaders ? or vertex ? or ... ? that makes the folding more efficient for a graphic card ? How about processors ?

It could be good to find the referential that can be tested to compare products, and generate benchmarks which everyone can see to make his own opinion.

Tom's hardware is a good site for benchmarks, and they have just published this article http://www.tomshardware.co.uk/tom-s-har ... 30644.html

I hope it will be possible to compare products for folding... someday...

viewtopic.php?f=52&t=3193

viewtopic.php?f=51&t=4263

There is

Shortcut -> ftp://legoman666.dyndns.org:47024/scores_nvidia.zip ftp://legoman666.dyndns.org:47024/scores_ati.zip

Edit: hehe if my english wasn't abit sloppy to say the least I would offer to write something about folding for them

I already contributed to get valid FAH benchmarks on the GTX 2xx release

That was for the French version of the site (yes, I know some people there) : http://www.presence-pc.com/tests/GeForc ... -22792/25/ (scroll down to the Folding@Home part, and you'll see the results ... that was with the p5000, the first project released for NV GPUs).

For technical dullards like myself do those links show Nvidia cards folding 3-4x faster than ATI cards of the same generations?
Is this because NVidia have worked closer with the Pande group than AMD/ATI or just the technical difference in the cards? ... I guess I'm asking will it likely balance out in the future and is there a point where the card utilisation plateaus ?

This has been discussed many times in other threads. (Search is your friend.)

The early WUs for NV were small test proteins whereas ATI was already working on production sized proteins. NVidia does perform exceptionall well on small proteins. They're still in transition toward production proteins and there are a lot of gripes that the NV points have gone down. The early ATI proteins depended on a lot of support from the CPU and the benchmark machine turned out to be CPU-limited, which also distorted the early comparisons.

Concurrent with all of that, both NV and ATI have been optimizing their drivers, improving both speed and stability. The final figures are not yet in.

That makes a bit more sense. I was just looking through some posts on the FAH Hardware side of the forum but as a layperson it all ends up looking like gobledigook Anyhow... will move my queries to more appropriate threads, cheers Bruce.

Thanks everyone