Message boards : Graphics cards (GPUs) : Short vs Long unit size
| Author | Message |
|---|---|
|
popandbob Send message Joined: 18 Jul 07 Posts: 67 Credit: 41,551,724 RAC: 84,863 Level  Scientific publications     | |
|
I was running short units in 3 hours with the CMYBKIX tasks on my 650Ti so I thought I must be able to run the long units in under 24 hours... boy was I wrong! 35 hours in and only 79% finished. Whoops guess I learnt my lesson! | |
ID: 37547 | Rating: 0 | rate:
 /   Reply
Quote
| |
|
mikey Send message Joined: 2 Jan 09 Posts: 298 Credit: 6,653,545,497 RAC: 15,025,234 Level  Scientific publications   | |
I was running short units in 3 hours with the CMYBKIX tasks on my 650Ti so I thought I must be able to run the long units in under 24 hours... boy was I wrong! 35 hours in and only 79% finished. Whoops guess I learnt my lesson! Two things that may or may not help...your version of Boinc is not the latest and neither is the driver versions for you gpu. One last thing to try is to leave one cpu core free just to keep the gpu fed and running optimally. I have a 760 and am doing long units in just under 10 hours normally, I would think your 650 would be less then 24 hours. | |
| ID: 37548 | Rating: 0 | rate:
/ Reply
Quote
| |
|
popandbob Send message Joined: 18 Jul 07 Posts: 67 Credit: 41,551,724 RAC: 84,863 Level Scientific publications | |
|
I was using the older drivers for folding@home performance... Guess now I'm running here again I can update. | |
| ID: 37549 | Rating: 0 | rate:
/ Reply
Quote
| |
 skgivenVolunteer moderator Volunteer tester  Send message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level  Scientific publications  | |
|
Excluding aforementioned observations, | |
| ID: 37550 | Rating: 0 | rate:
/ Reply
Quote
| |
|
I'm finding crunching short work units to be worthless on my GTX 560 Ti. It takes exactly double the amount of time to crunch a short work unit as a long work unit, but rewards are much higher: 130,000 vs 30,000. That said, it only takes ~17 hours to complete a long unit here, versus the 8.5 hours for short units. | |
| ID: 38384 | Rating: 0 | rate:
/ Reply
Quote
| |
Beyond Send message Joined: 23 Nov 08 Posts: 1112 Credit: 6,162,416,256 RAC: 0 Level Scientific publications | |
Q6600 - probably a DDR2 system (10 to 20% difference)? Interesting question and one for I have some experience to share. I'm now running 10 750Ti cards, all but one at about the same settings. Eight of the machines are running Phenom X6 processors, 6 with DDR2 and 2 with DDR3. The DDR2 machines are every bit as fast at GPUGRID (and other GPU projects) as the newer DDR3 boxes. They are also just as fast when running CPU WUs for various projects. Recently someone gave me a DDR2 Q6600 system and I've run a couple different cards on it (right now a 650Ti). The MB is of lower quality than the ones on the Phenoms (but is still PCIe2x16). It runs a bit slower than the Phenom systems for GPU WUs and quite a bit slower for CPU WUs but still draws as much power running 3x WUs as the Phenoms do running 5x. Also have 2 dual core Celeron 1037U ITX machines that run the 750Ti cards just as well as the above Phenoms X6s (but only 1 PCIe slot per box). The processors are pretty slow for CPU work but with 1 GPUGRID WU and 1 CPU WU draw only 87 watts (gold rated PS). As a reference a Phenom box running 2x 750TI cards on GPUGRID and 5x CPU WUs draws 281 watts (gold rated PS). Running in a PCIe2x4 slot however does slow GPUGRID WUs somewhat compared to a x16 slot but not more than a few percent (probably would be more with higher end GPUs). I doubt if there would be any difference at all between an x16 and x8 slot. | |
| ID: 38406 | Rating: 0 | rate:
/ Reply
Quote
| |
Message boards : Graphics cards (GPUs) : Short vs Long unit size
