Page 2 of 2

Re: Fair distribution for ppd BigAdv / Gpu

Posted: Mon Jul 04, 2011 1:53 am
by GreyWhiskers
I've been playing with another spreadsheet to look at possible QRB parameters for the GPU.

I've pasted a screenshot of the spreadsheet I'm working on - looking at varying the k-factor, the deadline, and a cap on the QRB bonus.

I'm not too sure about the cap - I've taken it as maxing out the number of points as the (base + base*cap). If the cap is 5, as in my example, the max total points is 6x the base. Without the cap, some of the faster numbers came out with ridiculously huge ppd values, considering the cost of the hardware.

I was just looking at what the numbers could be for a p6801 at 1348 points.

One thing I don't know is the range of times that work units like this get from the various folders. I've circled the region that I get on my 950 GHz GTX 560Ti (the TPF is a pretty consistent 1:21). I know that some of the 590s or corresponding ATI cards get much better performance than this. What's the cutoff point?

And, most important, what behavior on the part of the folding community do we intend to motivate by adding a QRB?

Thanks - another interesting exercise proving that I really need to get a life. GW.

Image