• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

NV GPU2 client - suspected first numbers

As someone on that forum mentioned, there's an enormous range of results from the stats. It's hard to say what to make of it besides the general level of production being very high. It bodes well for the NVidia client. Too bad I don't currently have a compatible card. :(


 
Wow. Some of the post in that thread are priceless. :D:D

I could see the 280 doing 3K PPD but certainly not much more than that. As for the 8800s - I dont thinkg they will be doing "3 -5K PPD".:rolleyes:

I am excired about this client for many of the same reasons why the PS3 client seems to do so well. That being that there are so many Nvidia cards out there and it will give many more people a chance to join the folding project. Whether it will be the most powerful, power-efficient, biggest-PPD client available only time will tell.
 
If the GPU2 client can produce over 2500 ppd, I can see it as a viable alternative to 2x SMP by setting SMP on 3 cores and GPU2 on the 4th or even SMP on 2 cores, GPU1 on the 3rd and console on the 4th.

IMHO, 2500 ppd would be the break point in favor of GPU2 but don't put all eggs in the basket and buy a gpu just for folding. If you need it for gaming, go ahead ;)

 
I've seen a GTX 280 put out 5978 and 4214 in one day, using 1 core on a stock clocked QX9770.
 
Xilikon said:
If the GPU2 client can produce over 2500 ppd, I can see it as a viable alternative to 2x SMP by setting SMP on 3 cores and GPU2 on the 4th or even SMP on 2 cores, GPU1 on the 3rd and console on the 4th.

IMHO, 2500 ppd would be the break point in favor of GPU2 but don't put all eggs in the basket and buy a gpu just for folding. If you need it for gaming, go ahead ;)

Click to expand...

From my experiences, it would be practically useless to run dual SMPs or even a single SMP instance on three cores. The client does not scale worth a damn using an odd number of cores. Basically, it needs even number of cores feeding each other. If it does not have this, one of the cores will sit there practically unused. Unless Stanford does something to the client or the core files, it's not something which will change anytime soon.

For the best production on a quad it would require the GPU client using one core, the standard client using one core and the SMP client on the remaining two cores. I would hate to see the possible conflicts this could cause with running that many different clients at once as well as the swapping between cores without something to automatically assign affinity to each client.

 
This is where running a VM for the SMP will help. It would take 2 cores, leaving you with the remaining 2 and you can assign the console client to the 3rd and the GPU will just pick the 4th (dunno if leaving both unassigned will work ok, more testing is required).



 
Has anybody seen any reports on how much extra power these cards use when folding.

At the moment my farms runs at about 22 Points per Day per Watt.
My quads do ~4,000 PpD for 176 watts drawn from the wall.
Adding a 125 watt vid card means the client would need to do ~4,500 PpD to break even in terms of PpDpWatt.
I dont see that happening.
I guess it will be benched to give around 3,000 PpD.

Luck ............. :D

 
Lord von Schtupp said:
I doubt the durability of the stock cooling (of the reference cards at least) for 24/7 folding.
Click to expand...

Yes, 85C is getting a bit hot especially for 24/7 use. I'm almost more interested to see how the new ATI cards are since they are supposed to be lean and mean. Unfortunately it seems there is always pre-release hype for the ATI side that fizzles after the cards are actually released.
 
As far as GPU folding is concerned looks like ATI would have the edge with the 55nm die, they certainly take less power at full speed, probably get a better watt/point ratio than the nvidia cards.
 
It should be rather interesting once this enters the realm of public consumption.

I have quite a few cards sitting idle, but if nvidia has the same fatal flaw as ati (needing a supercharged CPU) then i really don't see the point.
 
IcedEmotion said:
I'm almost more interested to see how the new ATI cards are since they are supposed to be lean and mean. Unfortunately it seems there is always pre-release hype for the ATI side that fizzles after the cards are actually released.
Click to expand...
There's been some speculation that the 4000-series cards perform lower than the latest NVidia cards in F@H, perhaps 1000-15000PPD lower. Even if they consume a bit less power, it could prove better to purchase NVidia on a point per watt basis. A point per dollar basis might be a totally different story.

 
Tigerbiten said:
Has anybody seen any reports on how much extra power these cards use when folding.

At the moment my farms runs at about 22 Points per Day per Watt.
My quads do ~4,000 PpD for 176 watts drawn from the wall.
Adding a 125 watt vid card means the client would need to do ~4,500 PpD to break even in terms of PpDpWatt.
I dont see that happening.
I guess it will be benched to give around 3,000 PpD.

Luck ............. :D

Click to expand...

For dedicated boxen and farming, adding a vid card probably wouldn't be the best course of action like you say. But for some of us that only have a couple boxen and are already powering an 8 or 9 series card, I think it could be a great addition.

Really, the only way I won't run the gpu client is if the 8800GT lowers my overall PPD from my main machine. But if what i have read is true, like on vijay's blog, they're trying to even the points out so if you've got the card already, it should be helpful.

Of course we won't know till it comes out. I think that it will depend on a couple of factors.
1, if you're running a dual core cpu, will the gpu you've got have enough horsepower to put out more PPD than the dual core did?
2. same as 1, but with a quad cpu. I think the quad's boxen will see a higher overall ppd with a midrange to upper gpu since you'll still be rolling an SMP client as well.
3. Model and clockspeed of gpu. Will the lower sp count of the 9600GT make it a worse performer than a 8800GTS 320 or the 8800GS/9600GSO? Will higher clocked gpu's like the 8800GT/GTS g92 and 9600GT trump the older 8800 cards? Will overclocking show tangible improvement in times like it does for cpu overclocking?
4. Lastly, can your cooling deal with all that extra heat this summer?

Obviously this is just me thinking out loud, feel free to correct me :)


 
I have a water-cooled 8800gts 320 that's never been used...can't wait to tinker with this one ;)
 
I have my gaming rig with twin 8800GTX's watercooled with a PA120.3 with 3 medium flow 48cfm fans atm... this project may require me to bump those up to high flows at 78CFM... Temps shall not go above 60c... Right now they hit about 45-50 depending on the inside temp. :)

This may require me to move my gaming rig into my folding room and invest in some extension cables for the keyboard, mouse and video! (Hell I've been thinking of that for a while anyways... just needed a good reason to do it!)



 
You must log in or register to reply here.
Back
Top