be a good idea to use smp-8 rather than smp-9
odd numbers of cores has a tendency to crash on some SMP WU's

ramcharger89

I have 2 GTX 460's running getting about 13,000 ppd per gpu with the new wu 7663.
Which is about 2000 points less per GPU running non beta wu's.

Fermis did not fare well on this core. Kepler got a big boost. I took my 550ti back to non beta and kept the 590 on beta for a modest increase in points. Much less than what I was expecting. Next move is a better driver from Nvidia to keep fermis in the game. They are power hungry and points per watt is much better on kepler.

I'll try SMP-8 and see how that goes Xavier, so far 9 hasn't been giving me problems.. I skimmed the folding forum and read something about using SMP 9 if 10 wasn't helping when folding SMP+The GPUs, but reading while tired is never a good thing :p.

Never mind, I got it working, working my 1st 7663 on the 660ti FTW in Linux!.

Switched 2 680s and 1 660 ti...
 
I feel a smile creeping in...

Looks to me like OCing to shave seconds off will make a major difference in QRB, i'd be interested to see the effects on larger WU's 
 
 
for example AB's 680's are
TPF 1.48 = 73165 PPD
my 690 are folding each GPU
TPF 1.41 = 79137 PPD
 
But as a side note. The base points are much lower. so if you are not folding 24/7 on your GPU and you pause one to go to work.  The points on the paused unit are going to be less then the old GPU WU by a lot.
 
This WILL be an issue because people will tend to want to cancel the partially done job for the much larger point return on a non paused job.  Just saying.

To clarify a little about the QRB i see that is going to be an issue and its something someone should relay to the beta team.
 
 
The current 3 hour WU's have a base pay out around 4k per WU with no QRB.. I can pause the WU. go to work come home fire it back up finish the WU still get my 4 k for the WU.
 
The Beta WU's have a much lower base of 1600. at TPF 1.41 =  it will pay out well over 9k per WU. Cool right?
 
But if i get 2.5 hours into the beta WU. pause it to work. When I come home from work and restart it. I will get less then 3 k for the WU if its almost done.
 if you are just in the very beginning say only 5 or 10% into the WU the payout tends to be less then 2 k. People are going to just cancel it out and get a new one. 
 
The QRB is pretty dramatic by shaving seconds but if you pause to game its going to be a major hit thats going to end up with lots of aborted WU's.
 
for example AB's 680's are 
TPF 1.48 = 73165 PPD 
my 690 are folding each GPU 
TPF 1.41 = 79137 PPD 
harrier just posted his 660 TI
TPF 1.59 = 62783 PPD
 
all great numbers if your never going to use your PC.
 
 
 
 

big GPU is going to be different that's a flag you set. I am looking at the general public user that's on a gaming box or have any plans at all to use the machine for other then folding.
 
The loss is more along the lines of 9 k for 3 hours or 2 k for 3 hours if you pause for work. its not a minor deviation over an 10 hour period.
 
If you make it too much a loss or a deterrent to actually use your own machine people will either A) not bother or B) abort low paying WU's.  its something they really need to take into concideration. saying wake early and pause it is fine but its not always going to happen.
 
BigAdvs is different in that you know your dedicating it to fold.
 
 
Here is another example of how quickly the QRB degrades on the smaller WU's. I understand its beta that's why i am trying to give feedback.
 
When i posted around an hour and a half ago, i had paused one of my GPU's on my 690's so i can use my PC. its value has dropped by 3,000 points in an hour and a half. That much point degrading over such a short period of time, people will be dumping WU's. I can see this coming from a mile away.

That's were the finish comes in handy, if you think your going to use the PC always set it to finish once it starts a WU.
That way you  know in 3 hours its done and doesn't pick up another WU. Then you game get ready for bed hit fold.
You sleep 8 hours worked 2+ WU hit finish and it finished before you hit the door to go to work, shut down PC wave bye bye cya when I get home.

then post it at Folding forums for V7 suggestions
(all honesty it's a great idea Jini)
  
cutoff time for getting last WU (scheduled Assignment)
 
let say start my computer (when I get home like 5pm) it starts with pickup of new WUs - it scheduled to stop picking them up at 4 or 5am ...3 hrs before I get up in the morning that way they all finish before I get up and you can turn it off before you leave
 
 

I know gamers. 
They will take more over less... every time. most people with high end GPU's  are your gamer's that's the market.
 
Windows scheduler can start and stop the F@H but its not going to do it logically. it will start and stop it when you tell it too. it will not care where a WU is when it does.
 
What I am saying is a little more complex.
 
If a WU will take you past 5 am, it is returned to the queue for reissue kinda logic. That way its not harming the project and not inconveniencing the folder. But that only solves one issue of being able to limit your folding day to a set time with out trashing the QRB.
 
The other is that when you install F@H as a non dedicated folder it pauses GPU folding AFTER it pulls a WU. So the times ticking right away.. It may be a good idea to make QRB a flag you set on or off for those WU's keeping static point WU's separate. that would be a solution for gamers and worker bee PC's

I wish they add a flag like -twounit and -threeunit and so on with summer coming on I can fold my GPU's at night with a 3hrs WU I could set -threeunits and it wouldn't pickup another after the 3rd.
That way I could start around 8:30 9pm and it do 3 WU by 6am and have it pause or shut down.
A client #finished WU shut down addition would also be a nice feather. Were it would shut down the PC completely
after x# are worked.

bowlinra

ramcharger89
I have 2 GTX 460's running getting about 13,000 ppd per gpu with the new wu 7663.
Which is about 2000 points less per GPU running non beta wu's.

Well 13K PPD is double what I'm seeing. How is are the GTX 470 doing?

The 470's are getting just over 30,250 ppd.

Excellent information Panther-X (Blue Ribboinish type of help IMO)!  I will give your settings info a try and would be happy to report back to you.  Where would you prefer that we provide you that feedback?  I'm a member of the FAH Forum too, so I can PM you or add a thread over there (whatever way you would prefer).
 
Thanks for this reply and for visiting with us today! 

Thanks for the tip! I'm set and testing as I type.
Name: max-units
Value: v='2'
Edit: hope that's correct, this is what's shown in log <max-units v='v=&apos;2&apos;'/> the command did.
Sad part is it picked up a 5hrs unit so it'll be 10hrs or so before I'll know if it worked..