• 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.

Will Fermi fold onceits released?

Vaulter98c

Vaulter98c

[H]ard|DCer of the Month - October 2009
2FA
Joined
May 21, 2008
Messages
5,840
OK, so I think I remember somehting about Fermi not working with the current generation of GPU2 folding, butthat it did work and infact GPU3 was semi designed around the new Fermi, but obviously GPU3 isnt here yet

For some back story, Im running Quad SLI (dual 295's) in Glados, and during the winter the heat kept the furnace off longer and could OC and get almost 35k just from these 2 cards. Then, the new series of WUs (you know the ones) came out and I had to drop to stock, and even at that I was around 100C. Then, they got even hotter, and 1 of my cards would almost always run at half speed (not 2D, just half speed) netting me a measly 22k. Now, with ambient rising, I hit 118 the other day so I shut all 4 down and only run 1 client per card at stock, givng me a pathetic (for the price) 14k.

Now I know the test on Fermi show what, 70% increase over a 285 (around 10k), on the GPU3 based benchmark, so the new Fermi that will take their place should be able to top my current summer levels output right? Even though I'm losing potential production, as long as these WUs stick around, I'm stuck at 13-14k on 700$ worth of video cards.

By selling them I have lowered my draw and my heat, and on the rare chance I game anymore, my Fermi will easily beat out a single GPU ( didnt enable QuadSLI for gaming vecause it screwed with folding afterwards) 275 pretty much. I'm just wondering if/when we will be able to fold on these bad boys, because I plan on getting on release day.

And if it can fold, you can bet your ass as soon as I get one I'll post some numbers :)
 
I never got the reason why when we have bigadv, people still go for 295s...?

You can get a Core i7 920 for $200, plus 6GB of RAM for $100, plus a cheap x58 mobo for $200, plus a cheap heatsink for $50, makes $550. This system would need about 100W less power than an individual GTX295 when overclocked (that's about $150 of power a year where I live if you run 24/7), doesn't cost that much more, and can put out about 30k ppd if you overclock (and this figure is pretty easy, given the OC potential of the D0s). So what gives?

The GTX480 will cost nearly $500, will go right up to the 300W limit in terms of power consumption, will run at 95C plus (on my 4870, F@H runs the core about 20C higher than anything else, even rthdribl), will last maybe a year or two at most running 24/7, and still may not even break 20k ppd depending on the GPU3 client.
 
Hence I just sold them, am getting one 470 (need a DX11 gamer, might as well get one that folds) and the remaining 250$ goes to buy ram for my dual xeon 5530 (already have them) SR-2 rig that should put out over 60k alone. In a few weeks, I'll only have 2 GPU clients running, a Fermi 470 and an 8800 GT for Physx

Everything else is optimized for power draw and heat, see sig :)

EDIT
and just to clarify, pushing an i7 folding is not like a normal OC, were finding about 3,8 to be the good spot before things start to fail, and that makes 25k, not 30

Both of our points are still valid, that bigadv is the way to go now, but I just want the record to be correct
 
I have no contacts or anything like that so take this info with a grain of salt.

What I've heard is that there will currently be no client for Fermi since it requires GPU3 to run. There is a beta client for Fermi and there is a chance that there may be an open beta for the GPU3 Fermi client sometime this month. However, I'm not sure if you should count on that. I personally don't expect there to be very many Fermi cards running around in the wild for a while yet and that means it could be a while before an open beta can be done for this client.

I would also caution you against expecting all that much in the way of performance from Fermi. Since it requires a new client and possibly new cores as well, I would guess that there won't be a whole lot of units for Fermi to run for a while. If new cores are not required and the older work units will run on Fermi, I wouldn't expect much performance out of them compared to current cards. We've already seen that there are a number of work units which do not run all that well on the high end nVidia hardware compared to the older hardware.

I would also take any performance numbers you've seen with the benchmark for Fermi with a large grain of salt. That will probably be the best case scenario with regards to Fermi. If you can even reach that level of performance, it will probably be limited to work units specifically made for Fermi. Since Fermi is a new architecture, there probably won't be a whole lot of work units put out specifically for it as the number of Fermi cards will be very low compared to previous generations of cards. Stanford cannot afford to put too much time and effort into work for Fermi when the vast majority of cards out there can't efficiently do the same work.

My guesses are based on the fact that Fermi is a completely different architecture and requires a new client just to run. This is very different from the transition from 8xxx series of cards to the 2xx series of cards which did not require anything like this.

 
Yes, Fermi introduces a new architecture. But the architecture is specifically designed to increase performance of such applications by making the GPU more CPU-like, which makes CUDA a lot more powerful; and after all, the first version of GPU3 will be on CUDA.
 
I just sold my 295 as well.

I will be buying a 470 once the Folding client for them is released. I don't have much gaming use for it, so I can wait until the GPU3 client rolls around. If anything, they will get cheaper, cooler, and faster in the meantime.

I know its pretty inexpensive to build another i7 machine now, but just dropping a card into an existing machine is much easier.
 
According to a post on FAH-Addict when Fermi came out, it should still be compatible with GPU2. I haven't seen anyone actually test that though.
 
Zero82z said:
According to a post on FAH-Addict when Fermi came out, it should still be compatible with GPU2. I haven't seen anyone actually test that though.
Click to expand...

It's not compatible. Kyle was going to fire up the two 480s he had but it wouldn't do anything. I also remember reading somewhere that Fermi does require GPU3 to run.
 
That's what I thought, I just hope there isn't much down time betweenr elease of Fermi and release of GPU3

but if there is, god forbid, I may actually get some gaming in on my gamer rig, blasphemy!! I don't even use Internet radio in the background for fear of losing ppd, I'm a changed man!
Posted via [H] Mobile Device
 
SmokeRngs said:
It's not compatible. Kyle was going to fire up the two 480s he had but it wouldn't do anything. I also remember reading somewhere that Fermi does require GPU3 to run.
Click to expand...
Kyle also didn't explain what he did or say whether or not he tried to use the -forcegpu nvidia_g80 flag. I haven't seen anyone who actually knows anything about folding really try to get GPU2 to run on a Fermi.
 
That's a good point, he didn't force the gpu, and it's supposed to be backwards compatible with older cuda based apps... Guess well find out in a week

anyone got a suggestion on which 470 to get? And FYI, I don't pay for OC's they are free, why buy 50mhz?
Posted via [H] Mobile Device
 
Vaulter98c said:
That's a good point, he didn't force the gpu, and it's supposed to be backwards compatible with older cuda based apps... Guess well find out in a week

anyone got a suggestion on which 470 to get? And FYI, I don't pay for OC's they are free, why buy 50mhz?
Posted via [H] Mobile Device
Click to expand...
I felt the same way until I bought two different brands of GTX275 - two EVGA and one ASUS
By default, the EVGA were clocked at 1512mhz shader, and the ASUS was clocked at 1420 ish shader

The ASUS is barely folding stable at 1512mhz shader

Both EVGAs are stable above 1600mhz shader, I think they're at 1620 and 1650 or whatever the next step is.

I thought it might be a fluke, but then I looked at the cards closely when they were out of my case, and I found there are some minor differences around the VRM area (can't remember specifics without looking again).

Thusly, I became slightly more interested in the "OC XXX SSC FTW" edition of video cards than I was previously
 
Ok, so when can I order these tonight? Who's midnight is the right one?
Posted via [H] Mobile Device
 
You can wait as long as you want, but you're probably going to be disappointed.
Click to expand...
Well, there's nothing I can do about it. So I might as well hope than be depressed.
 
newegg had pre orders? I have had their page up for almost 2 weeks, never saw that
 
My GTX 470 is en route. I'll let you all know if it can fold and how it goes when it gets here. Not interested in GTX 480 right now, it's over $540.00 Canadian. While I realize this is [H] I won't initially OC the card. Just not interested in another folding casualty from nVidia.

Apparently there is a bug. Hope it gets fixed by Thursday which is when my card will probably arrive from NCIX:

http://en.fah-addict.net/news/news-...ced-that-prevents-the-gtx4xx-from-folding.php

A couple of gents at EVGA had some success with it, but had a WU go corrupt at 83%

http://www.evga.com/forums/tm.aspx?m=293299
 
Damn! 10 minutes?? I figured this would be up atleast 2-3 hours, I woke up every hour a d a half to catch this, shit

Anywhere else got em? I'm gonna get on all the auto notify lists I can
Posted via [H] Mobile Device
 
Vaulter98c said:
Damn! 10 minutes?? I figured this would be up atleast 2-3 hours, I woke up every hour a d a half to catch this, shit

Anywhere else got em? I'm gonna get on all the auto notify lists I can
Posted via [H] Mobile Device
Click to expand...


the problem is the starting quantity is insanely low.. even worse then when the 5870 came out..
 
Last edited:
OK, they said they were in stock on MC.com, but I just got my backorder notice on the EVGA one from there. I think I got lucky tho and got an Asus from The Egg, I'm waiting on confirmation for that one now, if it comes in I'll cancel the MC order

too bad I'll be on duty all next week down in sharonville, but my new toys should be home when I get back!
Posted via [H] Mobile Device
 
Seems like it's a go, I just got charged by newegg, that means it's in stock right?
Posted via [H] Mobile Device
 
I have GTX 480 SLI on the way. They better fold! If not, gaming will be a beast!!!
 
You would think that this would force Stanford to get the move on GPU3
Posted via [H] Mobile Device
 
Ha! 1 of mine will arrive tomorrow! Too bad I'll be out of state until Thursday
Posted via [H] Mobile Device
 
DooKey said:
Can't read this at work. Cliff notes please.
Click to expand...

OK copied from the blog:-
"We have been working behind the scenes to optimize the Folding@home GPU client for the new NVIDIA GTX 4xx hardware. So far, it's been going well with us hitting some strong performance numbers. We are internally testing this and hope to soon (weeks) release this for outside beta testing. Please note that GTX 4xx support will require a new client and also requires some changes to our cluster backend software."

Not the news that you wanted Dookey i'm afraid:(
 
He said that fermi folding is working on their end, putting up good numbers, and in a few weeks they want it out to the public, so game on brother, because once it's here, no gaming lol

Also, he said it will require work on their platform and a new client for us, GPU3 maybe? Either way, they have it in alpha now, beta for the public (advmethods) soon
Posted via [H] Mobile Device
 
Nathan_P said:
OK copied from the blog:-
"We have been working behind the scenes to optimize the Folding@home GPU client for the new NVIDIA GTX 4xx hardware. So far, it's been going well with us hitting some strong performance numbers. We are internally testing this and hope to soon (weeks) release this for outside beta testing. Please note that GTX 4xx support will require a new client and also requires some changes to our cluster backend software."

Not the news that you wanted Dookey i'm afraid:(
Click to expand...


a few weeks for beta testing.. better not be the closed beta testing otherwise expect to wait months for it to come out..
 
when he says outside beta, I'm thinking advmethods
 
You must log in or register to reply here.
Back
Top