ppd Wu's/shaders

If you're new to FAH and need help getting started or you have very basic questions, start here.

Moderators: Site Moderators, FAHC Science Team

Post Reply
Ricorocks
Posts: 475
Joined: Thu Aug 04, 2016 1:49 pm
Location: Georgetown, Texas

ppd Wu's/shaders

Post by Ricorocks »

Ballpark excluding CPU, am I doing ok, or room to improve. Note Thanks to all Bruce, Foldy, & many others, things are much more stable now, from your guidance. :D

3 machines one with two gpu's, other two one gpu:

All 3 machines total 'shaders' = 3,712 <GTX's 960, 1060, 1050ti, 750ti>

AVG <no interruptions, stalled gpus etc> 750k ppd & 36 wu, in 24hr time span:

750,000ppd / 3712 = 202 ppd/shader

Does this make sense & is it good, bad, or ugly

Thanks
Nathan_P
Posts: 1180
Joined: Wed Apr 01, 2009 9:22 pm
Hardware configuration: Asus Z8NA D6C, 2 x5670@3.2 Ghz, , 12gb Ram, GTX 980ti, AX650 PSU, win 10 (daily use)

Asus Z87 WS, Xeon E3-1230L v3, 8gb ram, KFA GTX 1080, EVGA 750ti , AX760 PSU, Mint 18.2 OS

Not currently folding
Asus Z9PE- D8 WS, 2 E5-2665@2.3 Ghz, 16Gb 1.35v Ram, Ubuntu (Fold only)
Asus Z9PA, 2 Ivy 12 core, 16gb Ram, H folding appliance (fold only)
Location: Jersey, Channel islands

Re: ppd Wu's/shaders

Post by Nathan_P »

I've never worked it out based on PPD per shader, the usual metric is PPD/watt or PPD/$. The maths is right though.

As a baseline comparison my 1070 and 1080 are currently producing 1,500,000/4480 = 335 ppd/shader. My 980ti and 750ti are currently offline but would drag that down

All I would say is this, a pair of 1060's or 1070 will produce more points for less watts
Image
foldy
Posts: 2061
Joined: Sat Dec 01, 2012 3:43 pm
Hardware configuration: Folding@Home Client 7.6.13 (1 GPU slots)
Windows 7 64bit
Intel Core i5 2500k@4Ghz
Nvidia gtx 1080ti driver 441

Re: ppd Wu's/shaders

Post by foldy »

Looks good! The shaders of latest nvidia generation pascal are faster than previous one.
Ricorocks
Posts: 475
Joined: Thu Aug 04, 2016 1:49 pm
Location: Georgetown, Texas

Re: ppd Wu's/shaders

Post by Ricorocks »

Great! As noted, previously I'd have stalled GPU's routinely, sometimes multiple time per day. With your help it's very stable, rare GPU stall. So looking at the newer averages prompted the OP. PPD/watt or PPD/$ would also include the CPU's contribution, therefore better.

When I get my next machine (trade or freebie) I'll retire the 32 bit/750ti machine, & go 1070. I'll insure 'pascal on the new one

Thanks Guys
Nathan_P
Posts: 1180
Joined: Wed Apr 01, 2009 9:22 pm
Hardware configuration: Asus Z8NA D6C, 2 x5670@3.2 Ghz, , 12gb Ram, GTX 980ti, AX650 PSU, win 10 (daily use)

Asus Z87 WS, Xeon E3-1230L v3, 8gb ram, KFA GTX 1080, EVGA 750ti , AX760 PSU, Mint 18.2 OS

Not currently folding
Asus Z9PE- D8 WS, 2 E5-2665@2.3 Ghz, 16Gb 1.35v Ram, Ubuntu (Fold only)
Asus Z9PA, 2 Ivy 12 core, 16gb Ram, H folding appliance (fold only)
Location: Jersey, Channel islands

Re: ppd Wu's/shaders

Post by Nathan_P »

Not quite PPD/$or PPD watt is calculated one each platform separately, the calcs on GPU and CPU will lead to different results. I haven't run a cpu WU in several months due to the end of the "Bigadv" WU but both will be less than a gpu based on the current projects. Core A7 may change this but not while its stuck at 27 threads or less
Image
bruce
Posts: 20910
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: ppd Wu's/shaders

Post by bruce »

foldy wrote:Looks good! The shaders of latest nvidia generation pascal are faster than previous one.
When I compare GPUs, I've always multiplied the shader count by the frequency to create a number directly related to GFLOPS. It's still simple and it's a better approximation.
Post Reply