bit-tech.net

What is the Best Graphics Card for Folding 2011?

Comments 1 to 25 of 42

Reply
Spreadie 26th April 2011, 08:12 Quote
No surprises there then. Well, apart from the fact that a 560Ti can turn out roughly the same PPD as my GTX 295.

Still, it was a worthwhile update.

I wonder, what ever happened to the update that was supposed to improve performance on AMD cards?
confusis 26th April 2011, 09:20 Quote
Delayed as per usual..

Oh well - seems my GTS450 purchase tommorow is justified, shame I'm defecting to a smaller team..sorry guys!
borandi 26th April 2011, 10:00 Quote
Quote:
You'll really help the project (and therefore medical science) if you leave the folding client running 24\7

This project is not the whole of medical science. In fact, it barely hits the scanner. The randomisation implementation of the protein chains means it's a very brute force way of solving protein folding issues. If they actually delved deeper into the mathematics, they could find general results/equation-esque type expressions to solve a range of problems overnight, rather than several years plugging away using millions of Watts of power and energy.

People with ATI cards can help dramatically with other Distributed Computing projects, such as stellar dynamics (MilkyWay@Home, part of the BOINC network).
Toploaded 26th April 2011, 10:23 Quote
Please don't take this the wrong way anyone, I'm genuinely curious. What motivates people to spend the time, money on parts and electricity, to fold 24/7? Is it the warm feeling of helping the possible progression of mankind, or it it just a kudos thing? (such as Bit-Tech being proud of their top ten status). It does seem like a pretty cool collective project in all, and I'm impressed with how seriously some people take it. Just seems like such a selfless act I wondered what drove people into it.

I would expect, at the very least, Steam achievements for such an endeavor :)
GiantKiwi 26th April 2011, 10:24 Quote
@borandi: how much more efficient are the other projects for ATi Cards? mine are apparently barely pulling 3k ppd each according to that, which kinda seems pointless.
DbD 26th April 2011, 11:15 Quote
Quote:
Originally Posted by Spreadie
I wonder, what ever happened to the update that was supposed to improve performance on AMD cards?

It was never more then a hopeful rumour circulated by the fanboys.
impar 26th April 2011, 11:51 Quote
Greetings!
Quote:
Originally Posted by DbD
Quote:
Originally Posted by Spreadie

I wonder, what ever happened to the update that was supposed to improve performance on AMD cards?
It was never more then a hopeful rumour circulated by the fanboys.
Here you have a link for the fanboy rumour:
Quote:
F@H Core 16: Taking AMD GPU Folding to the Next Level?
...
http://img849.imageshack.us/img849/7704/71677314.jpg
...
So there you have it; the new AMD Core 16 project seems to do wonders for AMD’s Folding@Home performance. Currently there is only a single project making the rounds but there are surely more in the pipeline and what we’ve seen so far is only the first step down a very long road.
Spreadie 26th April 2011, 11:52 Quote
Quote:
Originally Posted by DbD
Quote:
Originally Posted by Spreadie
I wonder, what ever happened to the update that was supposed to improve performance on AMD cards?

It was never more then a hopeful rumour circulated by the fanboys.

Yeah, but an update that improves performance would only benefit F@H - so you have to wonder why they haven't produced something.
DwarfKiller 26th April 2011, 12:12 Quote
I get the feeling all the Radeon 5/6xxxs were working on Core11 Work Units.
I was running my 5870 through the BETA client and there is a huge difference in Folding performance between Core11 and Core16 units.
DbD 26th April 2011, 12:12 Quote
Quote:
Originally Posted by impar
Greetings!
Quote:
Originally Posted by DbD
Quote:
Originally Posted by Spreadie

I wonder, what ever happened to the update that was supposed to improve performance on AMD cards?
It was never more then a hopeful rumour circulated by the fanboys.
Here you have a link for the fanboy rumour:
Quote:
F@H Core 16: Taking AMD GPU Folding to the Next Level?
...
http://img849.imageshack.us/img849/7704/71677314.jpg
...
So there you have it; the new AMD Core 16 project seems to do wonders for AMD’s Folding@Home performance. Currently there is only a single project making the rounds but there are surely more in the pipeline and what we’ve seen so far is only the first step down a very long road.

I have every right to be very sceptical.

Since the 3xxx series of cards came out there have been threads full of fanboys proclaiming their cards are actually much faster then the green ones, quoting theoretical stats, and saying there's the magical update for folding home just around the corner that will unlock their true potential. Every new generation of card that comes out gets tested with folding home, is found to be slower, then gets followed by a rumour that in-fact folding home isn't using that new card properly and if it was performance would go through the roof. So far it's never happened, that new card never gets the *special* update, the next new card comes out and the cycle starts all over again.

[QUOTE=Spreadie][QUOTE=DbD]
Quote:
Originally Posted by Spreadie
Yeah, but an update that improves performance would only benefit F@H - so you have to wonder why they haven't produced something.

I can't believe ati wouldn't have done it already if it were even half way possible. Folding home gets reviewed everywhere, and has been for many years - it is the reviewers equivalent of quake 3 for gpu compute. Even if they ignored every other compute application out there, and just hacked together a special driver to make folding go fast it would be worth it for them. What conclusions can you come too? Either the ati engineers are incompetent, nvidia is in fact secretly paying of folding home to keep ati slow, or the ati cards just aren't very good at compute.
dunx 26th April 2011, 12:14 Quote
I know you guys push F@H but there are many other BOINC projects where an ATI card is far superior.
Could you do a rundown with those ? As I have 3 x GTX 460 OC's and would like to add an ATI 5870...

dunx
javaman 26th April 2011, 12:32 Quote
Quote:
Originally Posted by Toploaded
Please don't take this the wrong way anyone, I'm genuinely curious. What motivates people to spend the time, money on parts and electricity, to fold 24/7? Is it the warm feeling of helping the possible progression of mankind, or it it just a kudos thing? (such as Bit-Tech being proud of their top ten status). It does seem like a pretty cool collective project in all, and I'm impressed with how seriously some people take it. Just seems like such a selfless act I wondered what drove people into it.

I would expect, at the very least, Steam achievements for such an endeavor :)

Simple, the same reason anyone on here builds a £1000 watercooled super system. They love hardware! The other side of it is the amount of knowledge and fine tuning that goes into folding just to get the best out of the hardware. It also introduces problems which an everyday gaming machine doesn't always uncounter like running 4 closely spaced gpu's, which arn't designed to run 24/7. a few members have also lost someone to cancer or another condition folding is studying so in some ways this is their attempt at helping science progress to find a cure and better understanding how to treat patients. Personally I just fold for kudos as you put it, I'm skeptical if anything will come of it but am happy to try to get into the top 200 bit-tech folders. Its another achievement and folding is surprisingly addictive. If something good comes of it then it'll defo be worth it.there's also part of me feels that having hardware sitting idling when not gaming is a bigger waste than the electricity spent folding

As for an AMD update on their client. CustomPC did ask the question and due to resources they haven't looked at the client but will in the future. They are currently busy between their standard graphics drivers, physX alternative and pushing openCL for fusion. Eventually and hopefully if fusion takes off they will turn their attention to this popular yet niche past time.
Goty 26th April 2011, 14:18 Quote
Quote:
Originally Posted by GiantKiwi
@borandi: how much more efficient are the other projects for ATi Cards? mine are apparently barely pulling 3k ppd each according to that, which kinda seems pointless.

The only project I'm familiar with is MilkyWay@Home, but the results are basically the polar opposite of F@H; AMD cards lead the pack, with even a 4770 being faster than the GTX 480. The 5870 is roughly three times faster than the 480 on average in MW@H.
Christopher N. Lew 26th April 2011, 14:43 Quote
Thanks for publishing this - looks like I made the right decision to get more GTX560s (:-)
Enzo Matrix 26th April 2011, 15:09 Quote
Quote:
Originally Posted by DwarfKiller
I get the feeling all the Radeon 5/6xxxs were working on Core11 Work Units.
I was running my 5870 through the BETA client and there is a huge difference in Folding performance between Core11 and Core16 units.

Indeed. I get between 7K and 8K using my radeon 6950
Hakuren 26th April 2011, 15:28 Quote
Seriously. Considering AMD cards for F@H is simply bonkers. nVidia destroys every AMD imaginable in this competition. And it is no surprise that testing above prove the point. It is client architecture fault. AMD cards are not bad for sure, but without proper client software AMD computing potential is wasted in such applications.

And word to Bit-tech Crew. Guys you should really think about the title. The best means >the best<. The best does not allow compromises. And as such - for F@H - there is only competition between 2 GTX 590s and 4 GTX 580s for best possible solution or if someone want only one card then choice is obvious as even under-clocked GTX 590 outperforms GTX 580 by far. Like in sport. 1st is 1st and 2nd is nowhere. :)
Blarte 26th April 2011, 15:34 Quote
I have never folded a Graphics card, but would imagine ones with soft PCB's would be best.
Ta da ....
pass mi coat ..
dunx 26th April 2011, 15:41 Quote
I haven't fried one. But can't recommend that you eat them...

dunx

P.S. Hasn't school finished yet ? ; - )
Pete J 26th April 2011, 16:33 Quote
Quote:
Originally Posted by borandi
This project is not the whole of medical science. In fact, it barely hits the scanner. The randomisation implementation of the protein chains means it's a very brute force way of solving protein folding issues. If they actually delved deeper into the mathematics, they could find general results/equation-esque type expressions to solve a range of problems overnight, rather than several years plugging away using millions of Watts of power and energy.
Yes, I'm sure that probably some of the smartest people on the planet simply can't be bothered to do the maths and would rather just waste computer time :) .

Seriously though, the maths must be horrendous enough for them to use this method instead. Besides, they can still do maths in the meantime and if they're fortunate enough to find a solution, they'll have a shed load of data to allow comparison.
Madness_3d 26th April 2011, 17:10 Quote
Quote:
that can only use the GPU2 client such as the GeForce GTX 260 (rev2)

you mean cards that need the GPU3 client such as 5**...

And for an article the compares the merits of older cards you didn't throw many in there. A few people I know have farms of 9600GSO's or 8800GT's (because they're single slot etc) and I think It would have been useful if you had of included a few more legacy nvidia cards and a few less pointless ATI's. We all get it, they fold teh rubbish.
Barry_White 26th April 2011, 17:53 Quote
I Apoligise now for the following stupid question.

Looking at the "folding cost efficiency" graph a GTX460 1GB is at 65 ppd per £ and a 460 1GB produces around the 9000 ppd mark, surely that it does not just divide into each other otherwise that would be £138 a day? surely this can not be correct?
Xyllian 26th April 2011, 19:06 Quote
Releasing a article just after the new AMD client is released comparing nvidia and AMD and not using the new client?

Wow, just wow...
Bakes 26th April 2011, 19:27 Quote
Quote:
Originally Posted by Barry_White
I Apoligise now for the following stupid question.

Looking at the "folding cost efficiency" graph a GTX460 1GB is at 65 ppd per £ and a 460 1GB produces around the 9000 ppd mark, surely that it does not just divide into each other otherwise that would be £138 a day? surely this can not be correct?

It means PPD per £ as in - you will get 65ppd for the lifetime of the card for each £ you spend.

£138 is the price of the card to start with. It's not every day, it's 65 points per day per pound, not 65 points per pound per day, if you see what I mean.
Barry_White 26th April 2011, 20:08 Quote
Quote:
Originally Posted by Bakes
Quote:
Originally Posted by Barry_White
I Apoligise now for the following stupid question.

Looking at the "folding cost efficiency" graph a GTX460 1GB is at 65 ppd per £ and a 460 1GB produces around the 9000 ppd mark, surely that it does not just divide into each other otherwise that would be £138 a day? surely this can not be correct?

It means PPD per £ as in - you will get 65ppd for the lifetime of the card for each £ you spend.

£138 is the price of the card to start with. It's not every day, it's 65 points per day per pound, not 65 points per pound per day, if you see what I mean.

Now i do :) thanks for the answer.
Lenderz 26th April 2011, 21:51 Quote
Quote:
Originally Posted by Xyllian
Releasing a article just after the new AMD client is released comparing nvidia and AMD and not using the new client?

Wow, just wow...

That was my reaction, I was surprised Bit-Tech didn't even make mention of it.
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums