Page 15 of 15

Re: Nvidia Pascal GPU cards???

Posted: Wed Mar 01, 2017 12:06 pm
by Sven
Seems as if we have a new Folding-King.

Re: Nvidia Pascal GPU cards???

Posted: Wed Mar 01, 2017 12:54 pm
by foldy
The prices in nvidia store are quite linear:
1080 Ti $700 1100k PPD . 250W
1080 .. $550 . 850k PPD . 180W
1070 .. $400 . 600k PPD . 150W
1060 .. $300 . 350k PPD . 120W
So for each $150 more you get 250k PPD more.

Re: Nvidia Pascal GPU cards???

Posted: Wed Mar 01, 2017 3:30 pm
by Sven
Good comparison foldy :)

Re: Nvidia Pascal GPU cards???

Posted: Thu Mar 02, 2017 5:52 am
by FldngForGrandparents
I assume those estimates are on Windows? Or just base clocks? I get average 700k PPD on my 1070 FE. Will be glad it is causing the 1080 price to drop.

Re: Nvidia Pascal GPU cards???

Posted: Fri Mar 03, 2017 2:10 am
by kiore
FldngForGrandparents wrote:I assume those estimates are on Windows? Or just base clocks? I get average 700k PPD on my 1070 FE. Will be glad it is causing the 1080 price to drop.
Not sure about the OS, my GTX 1080 on win7 averages less that the 850kppd mentioned, more like 750-800k ppd.

Re: Nvidia Pascal GPU cards???

Posted: Fri Mar 03, 2017 2:36 am
by ComputerGenie
kiore wrote:Not sure about the OS, my GTX 1080 on win7 averages less that the 850kppd mentioned, more like 750-800k ppd.
What brand & clock is it? I'm getting 800-850 on my Win 7 card and it's pulling 3 monitors.

Re: Nvidia Pascal GPU cards???

Posted: Fri Mar 03, 2017 7:34 am
by rwh202
ComputerGenie wrote:
kiore wrote:Not sure about the OS, my GTX 1080 on win7 averages less that the 850kppd mentioned, more like 750-800k ppd.
What brand & clock is it? I'm getting 800-850 on my Win 7 card and it's pulling 3 monitors.
How are you getting your averages (and over what time period?)? I reckon on 800k for my 1080s. Looking at FAHClient, HFM and stats will all give different results. If using HFM, even averaging the PPD for 100 units will be misleading if you don't factor in run time as well. The best true measure is total points / total time as reported by stanford stats. Even then, if the total time is not long enough, the WU lottery will have a large effect.
To help make more meaningful comparisons, I think that FAHBench ns/day is probably more reliable, but it might be possible to convert that to an effective PPD measurement.

Re: Nvidia Pascal GPU cards???

Posted: Fri Mar 03, 2017 1:39 pm
by ComputerGenie
I get my averages by looking at "day after" points (how many points did rig x earn yesterday) and taking a literal average based on how many cards were running for how many hours. I feel that is the only way for any of us to get a legitimate PPD number to talk about. Because my Win 7 card gets stopped for personal/business use, actual math is the only way for me to get a quotable number for it.

Re: Nvidia Pascal GPU cards???

Posted: Sat Mar 04, 2017 5:52 am
by kiore
ComputerGenie wrote:
kiore wrote:Not sure about the OS, my GTX 1080 on win7 averages less that the 850kppd mentioned, more like 750-800k ppd.
What brand & clock is it? I'm getting 800-850 on my Win 7 card and it's pulling 3 monitors.
An EVGA hybrid, but really at this cutting edge some quite wild swings in PPD between units making it quite difficult to declare an average, I prefer to be conservate in my estimates but 600-900kppd is the general range so far.
The 1080ti promises much more capacity, I expect an even bigger range between work units with this model.

Re: Nvidia Pascal GPU cards???

Posted: Tue Apr 11, 2017 6:44 pm
by Duce H_K_
First Titan Xp tests. But not in F@H
Image
Expect ones