| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: The Clean Energy Project - Phase 2 Forum
Thread: problem with computing |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 22
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi, this project exited me and I decited to participate. But I'm having problems with computing.
----------------------------------------10 Days ago I received a task, which should take about 18 hours to finish. It's not problem for me, I use my computer many hours a day. But I was really pissed when after those 10 days it still got stuck at 3-4%. And I know it was somewhere between 30-40% yesterday evening (as well as many other days before). What's the matter? I must turn my computer off every night. I can't leave it running 24/7. Is there anything in my settings or something that I can change? I'd love to continue this project but I'm not willing to waste my power. Thanks for helping. Mike from Czech Republic [Edit 1 times, last edit by Former Member at Aug 12, 2012 11:46:52 AM] |
|||
|
||||
|
jonnieb-uk
Ace Cruncher England Joined: Nov 30, 2011 Post Count: 6105 Status: Offline Project Badges: 
|
You'll need to provide some more infomation before anyone can even begin to help!
----------------------------------------Check out this post in the FAQ for thesort of information required: http://www.worldcommunitygrid.org/forums/wcg/viewpostinthread?post=128396 |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
By the way, have tried to compute a few CEP2 WUs couple of days ago, but after 20 hours of computing got only 5 CPU hours.
----------------------------------------Now computing CFSW, there I get ~0.65 CPU hours for ~45 minutes of real-time computing. jonnieb-uk Too bad I czn't provide any technical info now =( [Edit 1 times, last edit by Former Member at Aug 12, 2012 12:07:29 PM] |
||
|
|
gb009761
Master Cruncher Scotland Joined: Apr 6, 2005 Post Count: 3010 Status: Offline Project Badges: 
|
Hi Mike, I'm assuming that you're attempting to soley crunch CEP2 Work Units. If so, what may be happening, is that the second step in the job (which is an extremely long one), may be starting from it's last saved checkpoint (i.e., after the first step) each and every time you start your computer up.
----------------------------------------This, unfortunately, is a real issue with CEP2 - as this extremely long step (often going for many, many hours - depending on your machine's capability), doesn't get a chance to checkpoint, and thus, move onto the next step when you restart your computer the next day. The question as to whether there can be a checkpoint on a more regular basis during this very long step, has been asked many-a-time, and unfortunately, due to the type of processing it's doing, it's not possible. What I would recommend, is to participate with other projects (most checkpoint every 15 mins or so, again, depending on your machine spec). Okay, it may not directly be contributing to the CEP2 project, but due to the load balancing that's going on in the background (where non exclusive crunchers process WU's that CEP2 crunchers aren't doing - because they're able to crunch CEP2 work units without interruption), in effect you are helping the CEP2 project. Hope this helps...  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Small correction: It's the third step that takes most time (job #2 in the result log, starting with job #0 and if fully computed, ending with step 16 aka job #15).
Volunteers that use the computer hibernate function (uses no power), works on Linux and Windows for most systems or standby (uses some power), do not incur progress loss. See the tips and explanation sheet http://cleanenergy.harvard.edu/documents/Tips_Tricks_CEP2_Custom_Settings.pdf , most importantly, those that have set their client to pause when using the computer [keyboard/mouse/joystick] do best to set "Leave Application in Memory when suspended" |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Dear KenR-cz,
First of all: Welcome, thanks for joining CEP, and we hope that you will stay with us for a while! Sekerob is absolutely right - due to the particular nature of the CEP calculations, the project very much relies on the use of LAIM. If you have to regularly shut down your computer, CEP may not be the right WCG project for you and you may be better off supporting one of the other great WCG projects. If you can make due with sleep or hibernation, you should be good to go. For details check out our Custom Settings recommendations linked in the footer. Best wishes from Your Harvard CEP team |
||
|
|
noderaser
Senior Cruncher United States Joined: Jun 6, 2006 Post Count: 297 Status: Offline Project Badges: 
|
Hibernation or "Deep Sleep" (OS X) should take care of these problems for you, writing the state of RAM to hard disk before shutting down. That way, you're not restarting from the last checkpoint over and over again--your computer will pick up where it left off when you put it into Hibernate. Sleep is another way to tackle this problem, but that won't do you much good if your computer loses power. Also, I often have difficulty getting computers to actually stay asleep.
---------------------------------------- |
||
|
|
rbotterb
Senior Cruncher United States Joined: Jul 21, 2005 Post Count: 401 Status: Offline Project Badges:
|
Thanks for this information on this post everyone. I've had similar issues from time to time with CEP2 WUs running for 8-10 hours during a day, then if I had any issue on my laptop with some application hitting a snag where I have to bounce my laptop, then on the next time back up, it seems the CEP2 WU is starting from scratch - probably lost in this long step. Now not all CEP2 WUs do this to me. Sometimes, it seems to be just fine picking up reasonably close to where it stopped. But I also seem to find that if there was a 'burp' or any issues on my laptop during the day, then I find while other WUs seem to pick up where they leave off, the CEP2 WU will generally drop all the way back to it started. At least now I know what is going on. I generally only run one or two CEP2 WUs per week nowadays anyway and the vast majority of the time they run to completion just fine, so I'll just keep crunching on and know that when my laptop has a bad hair day, that my CEP2 WU will just lose its work in this long step and that's life.
|
||
|
|
rbotterb
Senior Cruncher United States Joined: Jul 21, 2005 Post Count: 401 Status: Offline Project Badges:
|
One more question on a strategy of running this CEP2 more consistantly successfully - If I have a CEP2, and just stop it on weekends or during times I know it won't run long enough to get through this long step, but say resume it at the beginning of a workday where I know I'll have my laptop running a good 9-11 hours straight, should this consistantly allow these CEP2 WUs run long enough to get past this long step? A CEP 2 WU will generally run about 12 hours on my laptop, so even on one of these 9-11 hour workday sessions, I'll still have maybe 1-2 hours left of crunching in a second session - so if I get this far in, will it be safely past that long third step of the CEP WU processing?
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi rbotterb,
it's hard to give a general answer to this. A workunit consists of 16 individual jobs linked together in this sequence: 0 noopt.bp86.sto6g.n.sp -1 (very fast) 1 noopt.bp86.svp.n.sp 0 (fast) 2 bp86.svp.n.opt 1 (slow) 3 bp86.svp.n.bp86.svp.n.sp 2 (fast to very fast) 4 bp86.svp.n.b3lyp.svp.n.sp 3 (average) 5 bp86.svp.n.pbe0.svp.n.sp 4 (average to fast) 6 bp86.svp.n.bhandhlyp.svp.n.sp 5 (average to fast) 7 bp86.svp.n.m062x.svp.n.sp 6 (average to fast) 8 bp86.svp.n.hf.svp.n.sp 7 (average to fast) 9 bp86.svp.n.uhf.svp.n.sp 8 (average to slow) 10 bp86.svp.n.um062x.svp.n.sp 9 (average to fast) 11 bp86.svp.n.upbe0.svp.n.sp 10 (average to fast) 12 bp86.svp.n.bp86.tzvp.n.sp 3 (slow) 13 bp86.svp.n.b3lyp.tzvp.n.sp 12 (slow to average) 14 bp86.svp.n.pbe0.tzvp.n.sp 13 (slow to average) 15 bp86.svp.n.upbe0.tzvp.n.sp 14 (slow) The 2nd number specifies from which previous job it picks up an input, and in brackets I've indicated how fast each job is. Checkpoints are written after each job, but not within (the bottleneck step in each job is complicated to restart, and the solution we implemented turned out to lead to I/O problems; hence we had to make due without it). So if your computer is shut down you always fall back to the last checkpoint. The progress bar is only a rough measure and can be misleading. Hope this clarifies your question. Best wishes Your Harvard CEP team |
||
|
|
|