Nvidia Beta Tutorial(Early Return Bonus) for huge PPD

[Maban]

Move over to TPU and I will replace all your 4.8Mil points within a week under TPU. You 've been getting a steady string of 8057's?

I might switch over each month after I get my EVGA quota. I've been getting mostly 8057. I've only had a few others since I added the beta flag.

 

[johnspack]

s that right? I've only got my one Kepler going right now.

Yep, and my ancient 480 can pull 120k ppd with the beta units. Keplers don't have anything like that right now. They will soon I'm sure, but right now it's time for the fermi's to shine!

 

[Maban]

Yep, and my ancient 480 can pull 120k ppd with the beta units. Keplers don't have anything like that right now. They will soon I'm sure, but right now it's time for the fermi's to shine!

My 680 at +100 gets about 175k at about 120W over idle.

A month ago it took me an entire month to crack 1 million. I can get that done in 6 days now if I wanted with just SMP and one Kepler. Faster If I bring my Fermi's in.

 

[johnspack]

Well considering a 680 has more than double the horsepower of a 480, it better be able! When they come up with beta units for kepler though, bet you'll see 200k+ on a 680

 

[Feänor]

My 680 at +100 gets about 175k at about 120W over idle.

A month ago it took me an entire month to crack 1 million. I can get that done in 6 days now if I wanted with just SMP and one Kepler. Faster If I bring my Fermi's in.

The kepler core is not mature enough to fully utilize the 680, hence the minimal power consumption increase while folding.

I think, and correct me if i'm wrong, the gtx 580 have better double precision flops than the 680, so maybe even with a better folding core it will not surpass the fermi cards.

 

[Maban]

The kepler core is not mature enough to fully utilize the 680, hence the minimal power consumption increase while folding.

I think, and correct me if i'm wrong, the gtx 580 have better double precision flops than the 680, so maybe even with a better folding core it will not surpass the fermi cards.

F@H does not use FP64.

 

[NastyHabits]

Nice spike

I finally got V7 and the Beta folding squared away. Got a nice spike last night. One result greater than my average daily output. I'm running a conservative 840/1680 on my GTX 460.

 

[[Ion]]

Buck, might you consider updating the OP so it doesn't instruct new users to fold for OCN?

 

[BUCK NASTY]

Buck, might you consider updating the OP so it doesn't instruct new users to fold for OCN?

Done! Thanks for noticing that [Ion]. I will post a re-written tutorial with proprietary screenshots later tonight.

Wow, produced 310K PPD with only a GTX460, GTX560ti448, and a GTX570 yesterday. These P8057's are for real. I will admit that I was being a little selective.

 

[johnspack]

Heh, finally got an 8057. Turns out my os was unstable, so I did a fresh install just to test. Lost some ppd though because I didn't realize when it dls a beta wu for the first time, it creates a new beta folder, and dls the newer fahcore to that. So I replaced that fahcore and of course my ppd jumped up.

 

[BUCK NASTY]

It appears that new work unit's(P8057) are requiring a update to Fahcore 2.25. I cannot get any to run under 2.22. Can anyone confirm this?

 

[NastyHabits]

It appears that new work unit's(P8057) are requiring a update to Fahcore 2.25. I cannot get any to run under 2.22. Can anyone confirm this?

Don't know. Glad I didn't switch yet. I'm not getting the numbers others are getting on 8057's (just 70K PPD on my GTX 460), but still way more than before.

Also, pray you don't get a 2046. Took over 10 hours!

 

[BUCK NASTY]

Well, it appears to be real. Everybody is getting it and complaining as well. We should not complain, as we are still getting 4-5x the PPD we were getting before Early Return Bonus.

 

[mstenholm]

It appears that new work unit's(P8057) are requiring a update to Fahcore 2.25. I cannot get any to run under 2.22. Can anyone confirm this?

We should not complain, as we are still getting 4-5x the PPD we were getting before Early Return Bonus.

http://img.techpowerup.org/121204/Capture059.jpg

I of course didn't notice the difference since I ran 2.25 all the time (680) but I don't get 4-5 times, I get 8.8 times the PPD (185K for 8057 and 21K for 8054 and the like). So what kind of PPD do you get now on the 4 and 5xx serie?

 

[Caspase]

Sorry if this is the wrong thread to ask, but are you doing this for something like that folding@home stuff the ps3 had?

 

[BUCK NASTY]

I of course didn't notice the difference since I ran 2.25 all the time (680) but I don't get 4-5 times, I get 8.8 times the PPD (185K for 8057 and 21K for 8054 and the like). So what kind of PPD do you get now on the 4 and 5xx series?

2.25 core
GTX560Ti 448 - 108K PPD(166k Previously)
GTX570 - 111K PPD(180K Previously)

Still definitely worth folding with.

 

[mstenholm]

Sorry if this is the wrong thread to ask, but are you doing this for something like that folding@home stuff the ps3 had?

Yes and I think the word "had" is correct for ps3. GPUs and CPUs is still working so please join in. A good/easy start would be to download the client (https://fah-web.stanford.edu/projects/FAHClient/wiki/BetaRelease ) and FahControl (https://fah-web.stanford.edu/projects/FAHClient/wiki/FahControl ). TPU is team 50711. If you have a never NVidia GPU you want a passkey
for some extra points. There is a guide in our folding forum but just ask.

Edit: I see that you have a 670. You want a pass key http://fah-web.stanford.edu/cgi-bin/getpasskey.py

 

[mstenholm]

2.25 core
GTX560Ti 448 - 108K PPD(166k Previously)
GTX570 - 111K PPD(180K Previously)

Still definitely worth folding with.

Yikes, but still way past what we are used to.

 

[Caspase]

Why do I want a pass key exactly? Isn't the point of this to share information towards research and such? Yeah, I remember doing this on my ps3 early days... Sure, I'll join in!

 

[[Ion]]

2.25 core
GTX560Ti 448 - 108K PPD(166k Previously)
GTX570 - 111K PPD(180K Previously)

Still definitely worth folding with.

It seems to me like Stanford is really shooting itself in the foot by forcing out 2.25 on Fermi cards. It doesn't seem like such a move is in their best interest...

 

[mstenholm]

Why do I want a pass key exactly? Isn't the point of this to share information towards research and such? Yeah, I remember doing this on my ps3 early days... Sure, I'll join in!

In order to get bonus points for early return (for GPU only for 8057) you have to prove that your GPU/CPU is returning good results. In fact you have to do 10 good ones before you can earn the bonus. The difference is in my case 2549 without bonus for 2 hours and 40 min or 20.600 with so you do want a passkey. OK the science completed is the same but ....

 

[NastyHabits]

So what kind of PPD do you get now on the 4 and 5xx serie?

I'm getting 6 times my best PPD on 8057 (70K vs. 11K) with my 460. I don't notice any difference on other projects. (I can't remember the exact project numbers).

 

[BUCK NASTY]

It seems to me like Stanford is really shooting itself in the foot by forcing out 2.25 on Fermi cards. It doesn't seem like such a move is in their best interest...

I know that there was a lot of flack from the 4P rig armies([H] & OCN) when Pande Group(PG) announced the GPU Early Return Bonus(ERB). There are even PM conversations that were posted between [H] members and one researcher @ Stanford concerning the upheaval in GPU points vs a 4P Opteron rig. This forced core is just a step in the process of assigning the correct amount of points for the GPU's and further unification of the V7 client. I don't see it dropping much more, as this would discourage all the new GPU folders that PG has picked up in the last month. It's a delicate balancing act now....

Those that have been folding for years are familiar with the Pande Group rollercoaster of hardware support/cores changes and the apparent lack of concern for the hardware donor. I believe the ERB was the first step of redemption for PG. We'll have to wait for the dust to settle, so don't make any major hardware purchases unless you can get rid of it easily(GTX6xx cards, etc). Still, these GPU points are great for now and I will keep my cards folding for a cure!

 

[mstenholm]

I
Those that have been folding for years are familiar with the Pande Group rollercoaster of hardware support/cores changes and the apparent lack of concern for the hardware donor..

Still, these GPU points are great for now and I will keep my cards folding for a cure!

Da, evet, ja, si, yes and what ever it is call in Thai (my next job so I better learn) to both statements.

 

[mstenholm]

"beta-testers" better read this http://foldingforum.org/viewtopic.php?f=85&t=23101&p=230931#p230929. I had WU stuck this morning and had to deleted the work-folder and judging from team output I don't think that I'm the only one with that problem.. Got a 7624 with projected PPD of 44K. Much better then then 8054 but far away from the 8057.