PrimeGrid
Please visit donation page to help the project cover running costs for this month

Toggle Menu

Join PrimeGrid

Returning Participants

Community

Leader Boards

Results

Other

drummers-lowrise
1) Message boards : Problems and Help : All wu's on a box errored at the same time. (Message 85770)
Posted 1836 days ago by Mark
No 4 cpu tasks & 2 GPU tasks.
TRP
ESP
PSP
SOB
2x GNF WR

All errored at the exact same second. I was asleep at the time, nothing else should have been running.
2) Message boards : Problems and Help : All wu's on a box errored at the same time. (Message 85765)
Posted 1836 days ago by Mark
All of my running wu's on one machine (#473510) errored at 24 May 2015 | 6:15:12 UTC. Windows event viewer doesn't show anything happening at that time.
Any ideas as to what would cause 6 different tasks (only 2 in common were WR Genefers on the GPUs) to all fail at the same time?

On the plus side the low powered crunching box I recently built had a PPSE prime found overnight.
3) Message boards : Problems and Help : GTX660 continually has genefer errors, but sieves all day long (Message 84938)
Posted 1873 days ago by Mark
I have a MSI Twin Frozr GTX 660 in a box that is currently dedicated to crunching. Genefer, both short and WR units always crash out (at various points from 12-94%) with a computation error. I have tried down clocking the card as far as EVGA Precision will allow me to but it has not helped.
It will sieve just fine, and that's great, but for events like the upcoming challenge it would be nice not to let a GPU go to waste. Are there any other solutions I can try other than down clocking the card? Its not a heat issue, it has never gone above 51C.
4) Message boards : Problems and Help : Is this a fluke? (Message 63121)
Posted 2647 days ago by Mark
I have received almost exclusively Cullen WUs for the last week, with a few SGS, PPS, and a PSP thrown in the mix.

Its not a problem per se, since its quickly working its way to silver, but I used to get a nice mix of projects.
5) Message boards : Number crunching : Valentine's Day Challenge (Message 62626)
Posted 2665 days ago by Mark
Ok, I have tasks now. Weird. Wish I had not lost 6 hours on the PC though :(
6) Message boards : Number crunching : Valentine's Day Challenge (Message 62624)
Posted 2665 days ago by Mark

I was getting CPU PPS (seive) tasks and PPS (llr) tasks, even though those were not selected on preferences.


I am now getting CPU PPS (sieve) tasks again, and preferences show that that is supposed to be GPU only. I get these tasks once every couple of weeks and I am not sure why. I especially do not want it during a SG event, when I can't get SG tasks.
7) Message boards : Number crunching : Valentine's Day Challenge (Message 62623)
Posted 2665 days ago by Mark
I have a PC that is not getting any SG work units. I have restarted BOINC, I have rebooted. I have selected GFN and SG only to try to force some tasks. I cannot get *any* tasks for this PC.

I was getting CPU PPS (seive) tasks and PPS (llr) tasks, even though those were not selected on preferences.

I aborted those, and now I cannot get any tasks.

Any ideas?
8) Message boards : Generalized Fermat Prime Search : Are the world record WU touchier then the short ones? (Message 61893)
Posted 2688 days ago by Mark
Your errors back in December were most likely caused by overclocking, but the one recent error was not. The error you got there was likely caused by insufficient available video memory following a restart of GeneferCUDA.


I got another one of these errors after about 24 hours or crunching. The short tasks give me no problems. Is there way to try to avoid these errors?
I would like to crunch some world record WUs, but will be happy with the short tasks if that is all that I can crunch for some reason.

Also, the GPU was running 4C cooler with this WU than the previous one if that would have made any difference.
9) Message boards : Generalized Fermat Prime Search : Are the world record WU touchier then the short ones? (Message 61834)
Posted 2691 days ago by Mark
Judging work by temps since they all show 100% CPU usage, those seem significantly less taxing. Hopefully I won't run into an error again.


It shouldn't make a big difference whether you're crunching a large number (SoB) or a small number (SGS) -- LLR-AVX is going to run about as hot as any program can run. (I could actually make an argument why crunching smaller numbers would run hotter than crunching larger numbers under certain conditions.)



When I was running the Proth & Sophie Germain llrs (not seive, the only CPU running sieve is the AMD), I was getting CPU temps ranging from 57-62C. The SOB with all 4 cores was running 67-72C.

I do feel pretty confident in this OC as I have been running WUs nonstop since I broke my arm and everything is stable and cool (except for the SOBS).
10) Message boards : Generalized Fermat Prime Search : Are the world record WU touchier then the short ones? (Message 61830)
Posted 2691 days ago by Mark
I am going on a small limb here with the comments already made. You are running some very large numbers on the processors at the same time as a WR. The error was a shared memory error. You are using all your processors (4) on the box. GPU likes to have a little bit of a CPU every so often and it may have tried to make a CPU call at a time the other 4 units were in an intense spot of calculation and refused to give up their spot.


It appears to me that I am using 50% of the processors. The SOBs, running on all 4 cores were pushing temps into the mid 70's. I dropped to 50% CPU usage and now the CPU temps are ok.

It sounds like this may have been "just one of those things". After I work through the SOBs, I may change to just running PPS & SGS WUs on the CPU and try a WR Genefer WU.

Judging work by temps since they all show 100% CPU usage, those seem significantly less taxing. Hopefully I won't run into an error again.


Next 10 posts
[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2020 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 2.34, 2.00, 1.61
Generated 3 Jun 2020 | 9:13:45 UTC