Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
The Bottom Line
A day late and a dollar short. Cliché but accurate. The Radeon HD 2900 XT is late to the party and unfortunately is bringing with it performance that cannot compete. The GeForce 8800 GTS 640 MB is $50 cheaper, performs better, and draws a lot less power than the 2900 XT.
This is as good as it is going to get for a while from ATI. The GeForce 8800 GTX will still dominate at the high end of the video card market. Of course we do not know about DX10 games yet, and there is no way to make any predictions how that comparison will turn out. As it stands right now the Radeon HD 2900 XT, in our opinion, is a flop. ATI needs to get its act together quickly. It needs to push out the mainstream cards soon and it needs to deliver a high end card that can actually compete at the high end of the market.

I read something similar in one of the reviews, which mentioned that a big design advantage of the R600's architecture was in scientific applications and distributed computing. Yes, the article actually mentioned specifically distributed computing. So, whatever one's opinion of the R600 might be in other applications such as gaming, it shouldn't flop in DC. Irrespective of any subjective viewpoints, it remains to be seen until someone actually uses one with the F@H GPU client, providing the client even properly supports the new architecture at this time.No client out yet for the card, coming on the next release cycle (supports more scientific core types IE different fah_cores.exe). Currently ~ 45% increase over x1900 in overall production per Houston.
Well it might not flop in DC, it's still going to require power supply upgrades for most folders should they decide to go with it. ATI's calling for a minimum of something like 600 watts and it also has the new 8 pin PCIe power connector. Most of your folding farms don't have beefy power supplies in them. Personally, I'll probably wait for the 3 month respin at 65 nm that will bring ATI up to speed with GTXs if not surpass them. This card will end up as a stop gap, just as the x1800 did.
I largely agree, but it all depends what system one has at the present. If you're running an AMD system, changing a motherboard+CPU+RAM (if necessary) and reinstalling the OS with all the apps might not be worth it for some, both on a monetary and/or convenience level. It really depends where you're coming from. If you're looking at a new video card that has decent gaming performance and the ability to run F@H, which is my situation, it might not be a bad solution once the price drops or the 65nm version with (hopefully) much better performance arrives.From the PPD viewpoint, I still believe a C2D setup is a better investment than getting this card.
From the PPD viewpoint, I still believe a C2D setup is a better investment than getting this card.
Depends on the new client. They're expecting some changes in points for the SMP and GPU WUs once version 6 is released. Depending on the significance of these changes, it may turn out in your favor to run a GPU client.
45% faster is substantial enough, I can only imagine double or greater performance. What would that bring it in terms of average minutes per frame on a typical WU, anyone can hazard a guess?mhouston has stated it has potential for 2.2X performance compared to the X1950XTX at F@H.. but currently they have it running at 45% faster due to driver immaturaty, cpu overhead and some other factors but the hope to bring the performance up substantially.
45% faster is substantial enough, I can only imagine double or greater performance. What would that bring it in terms of average minutes per frame on a typical WU, anyone can hazard a guess?
The other downside is that it isn't going to be a $400 upgrade for most folks. You are also going to have to invest $150 in a power supply that can handle the draw which puts you above the cost of a C2D complete PC from Dell or somewhere else that will draw less power and give you more PPD than the 2900 will from the get go.
In light of the estimated PPD, I will stand with this opinion that a C2D setup is cheaper.
Easy combo for 400$ = C2D E4300 + DS3 + 1 GB of RAM + Fortron PSU
frame times on my x1900xtx are just a tick over 6 minutes, say 6 minutes 12 seconds or so, about 780 PPD, so increase the PPD by 45% and that would be approx 1150 ppd. Still less than a good C2D setup. I was really dissappointed with the 2900 results, I was looking at one for folding and gaming, now it looks like folding is it's real strong suitethe good news is Crossfire just came down a little for me, bad news, crossfire=no folding
![]()
One thing to remember about the C2D setup is that one will rarely get the right combination of proteins to make 1000PPD+. I run a E6400 @ 3.0ghz, and typically produce 600-700PPD, and yes I have it set for Advmethods. I just rarely get those 2 juicy proteins at the same time.
Whereas the GPU client may produce a little less PPD, but it is much more consistent...
One thing to remember about the C2D setup is that one will rarely get the right combination of proteins to make 1000PPD+. I run a E6400 @ 3.0ghz, and typically produce 600-700PPD, and yes I have it set for Advmethods. I just rarely get those 2 juicy proteins at the same time.
Whereas the GPU client may produce a little less PPD, but it is much more consistent...
To get the 1K+ PPD folks are utilizing the SMP client not the standard console. My e6400 at 2.13 (I dont overclock) can chew through pretty much any SMP WU in ~24-48 hours depending on the WU. Since these are really skewed right now in the point system for example
Lelliot731 is putting out roughly 15K PPD and turning in on average about 21WU a day and seems to be pretty much SMP only
I am currently putting out roughly the exact same PPD but am turning in around 50WU a day because I use a LOT of console clients and not SMP.
He has 30 active CPUs and I have 121 active CPUs. I expect at some point for the points to balance themselves back out but right now the C2D with the SMP client is pretty much the best PPD that you can squeeze out of anything dollar for dollar.
That being said the science is much more flexible with the console client as the following holds true as far as I can tell
GPU Client - Fastest folding per core but very limited WU that can be done on it
PS3 Client - Second Fastest at folding but again can only utilize specific cores/WU
Win/Linux/Mac SMP - 3rd in line for speed, still limited in what WU it can do, best PPD for dollar spent
Console Client - Can do pretty much any of the WU but is significantly slower and pretty lousy PPD unless you can throw a LOT of cores at it (see FLECOM who runs all console as far as I can tell)
Graphical Client - For all things Holy don't use this![]()
Good question because I tried it on two Windows AMD machines and both crashed the client many times. I never completed a WU on either machine in the week and a half I attempted to run it. Mind you, they probably wouldn't have made the deadline, but that's another story.Did Stanford ever get the SMP for WINDOWS working? I refuse to screw around with Linux...
Good question because I tried it on two Windows AMD machines and both crashed the client many times. I never completed a WU on either machine in the week and a half I attempted to run it. Mind you, they probably wouldn't have made the deadline, but that's another story.
Did Stanford ever get the SMP for WINDOWS working? I refuse to screw around with Linux...
.......................
GPU Client - Fastest folding per core but very limited WU that can be done on it
PS3 Client - Second Fastest at folding but again can only utilize specific cores/WU
Win/Linux/Mac SMP - 3rd in line for speed, still limited in what WU it can do, best PPD for dollar spent
Console Client - Can do pretty much any of the WU but is significantly slower and pretty lousy PPD unless you can throw a LOT of cores at it (see FLECOM who runs all console as far as I can tell)
Graphical Client - For all things Holy don't use this![]()