Loudbob how is your general system performance running 3 WUs on each card? I tried running 3 on my 580 rig (2500k) and 470 rig (920) and both ran at 100% memory usage. The rigs became extremely unresponsive and couldn't be used for anything else.
SETI@Home & OcUk Projects News Vol. 174 7th April - 27th April 2012
- Thread starter ToxicTBag
- Start date
The 680's don't quite max out - so the rig can be used without any problems.
No idea if it makes any difference but the 990X is running @ 4.3GHz with 12Gig (2000MHz) ram.
The 2600K with the 560's runs fine too - CPU @ 4.5GHz with 8Gig (2133MHz) ram. Can even watch 1080p video without stopping seti - guess I'm just lucky. But most of the time it's just dedicated to seti or fah.
No idea if it makes any difference but the 990X is running @ 4.3GHz with 12Gig (2000MHz) ram.
The 2600K with the 560's runs fine too - CPU @ 4.5GHz with 8Gig (2133MHz) ram. Can even watch 1080p video without stopping seti - guess I'm just lucky. But most of the time it's just dedicated to seti or fah.
Sorry I wasn't being very clear, I was talking about GPU memory being maxed out 
System RAM has never been an issue.
My 580 rig is currently running 2 WUs per card. GPU Memory usage has peaked at 1389mb and these are only 1.5gb models. Having said that I have been able to play Trine 2 on maximum settings while seti is running and still got playable fps.
My 470 rig is at 1204mb GPU memory usage and they are 1.25gb cards.
There is no way either of my rigs can handle another WU and I do wonder how others manage it.
System RAM has never been an issue.My 580 rig is currently running 2 WUs per card. GPU Memory usage has peaked at 1389mb and these are only 1.5gb models. Having said that I have been able to play Trine 2 on maximum settings while seti is running and still got playable fps.
My 470 rig is at 1204mb GPU memory usage and they are 1.25gb cards.
There is no way either of my rigs can handle another WU and I do wonder how others manage it.
....for those interested in the S@H 'long game' this thread:
http://setiathome.berkeley.edu/forum_thread.php?id=67906
hints vey strongly at what you can expect from a 680
eek, using the optimised app' currently being beta tested (read the post from Jason G on May 2nd, 17:39 UTC). This app' he is referring to is a Fermi optimised app. A kepler optimised version is still on the back of his virtual fag packet. I know where I would put my money - if I had any 
!...and don't forget, a 680 produces less heat & uses less power than a 580, let alone a 480!
Last edited:
Sorry I wasn't being very clear, I was talking about GPU memory being maxed out
My 580 rig is currently running 2 WUs per card. GPU Memory usage has peaked at 1389mb and these are only 1.5gb models. Having said that I have been able to play Trine 2 on maximum settings while seti is running and still got playable fps.
My 470 rig is at 1204mb GPU memory usage and they are 1.25gb cards.
There is no way either of my rigs can handle another WU and I do wonder how others manage it.
My GPU's max out at around 900MB on the memory usage when running three wu's........but usualy stay around 700 - 750MB
He's obvously using the yet to be released Opt App.......
Each 680 averaging over 30k ppd
Also, Host ID "5631447" is using the Opt App on a single 680 and also averaging over 30k ppd.
Hopefully, it won't be too much longer before the app is released I then my 680's can get into it a bit more
Sorry I wasn't being very clear, I was talking about GPU memory being maxed out
My 580 rig is currently running 2 WUs per card. GPU Memory usage has peaked at 1389mb and these are only 1.5gb models. Having said that I have been able to play Trine 2 on maximum settings while seti is running and still got playable fps.
My 470 rig is at 1204mb GPU memory usage and they are 1.25gb cards.
There is no way either of my rigs can handle another WU and I do wonder how others manage it.
Make sure you disable sli.....
For some reason sli doesn't play well with seti.......
Make sure you disable sli.....
For some reason sli doesn't play well with seti.......
That did the trick, thank you good sir.
Good news......Bad News......
Bad News: Two GTX 560ti's (448 core) are now bosed up ready to return due to being unstable at stock settings
Good News: Two GTX 570's have now replaced them
Temps have also dropped by some 10 degrees on both GPU's and CPU
Just got to install a 460 now (in third PC to go with the other two 460's) - but its gonna take a little DIY to a spare mounting bracket which I will do at work tonight.......advantages to working in a rather large ship building facility
Bad News: Two GTX 560ti's (448 core) are now bosed up ready to return due to being unstable at stock settings
Good News: Two GTX 570's have now replaced them
Temps have also dropped by some 10 degrees on both GPU's and CPU
Just got to install a 460 now (in third PC to go with the other two 460's) - but its gonna take a little DIY to a spare mounting bracket which I will do at work tonight.......advantages to working in a rather large ship building facility
Last edited:
BTW....... I've nearly caught you
Yes I know. Less said about that the better
Think this will be my final day of glory, you big meany
Yes I know. Less said about that the better
Think this will be my final day of glory, you big meany
This is something I've been thinking about too.... since my 7970 never seems to max out even when I allocate 3 work units to it, but I was thinking maybe that's because I only left it at the default of 5% CPU per 0.33 GPU units?
This is how my GPU usage graph looks:
So I was wondering if there was any way I could eke out that last few percent. At first I was thinking it was just because the apps weren't yet optimised for the 7970 and I'd have to wait, but if other people have the same issue, maybe upping the amount of CPU percent dedicated to the GPU units would help?
Another thing I was thinking was, wonder what happens if in the app_info I allocate more than a whole GPU share and only have one GPU... like say if I allow 3 WU's per device, at 0.40 GPU share (a total of 1.2 GPU's) but only run one GPU? Will this max it out?
Last edited:
