| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Conquer Cancer
Thread: [RESOLVED] CPU WU shortage? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 6
|
|
| Author |
|
|
CandymanWCG
Senior Cruncher Romania Joined: Dec 20, 2010 Post Count: 421 Status: Offline Project Badges: 
|
I've been noticing intermittent issues in the past few days where, on my laptop that is only committed to HCC CPU WUs, I get the nasty "Tasks are committed to other platforms" message. Is it because most WUs are now for GPU or are there some issues at the back end?
----------------------------------------Thanks! Knowledge is limited. Imagination encircles the world! - Albert Einstein   [Edit 1 times, last edit by CandymanWCG at Mar 13, 2013 10:07:19 AM] |
|||
|
||||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
When too many repair jobs are queued, this happens [If your host is reliable, you'd be mostly likely getting rush job assignments]. There is *No* special HCC CPU queue. Tasks for CPU and GPU come from same feeder.
|
||
|
|
CandymanWCG
Senior Cruncher Romania Joined: Dec 20, 2010 Post Count: 421 Status: Offline Project Badges:
|
Thanks, SekeRob, but not sure I fully understand that. Indeed, my laptop is marked as reliable and I have been getting rush jobs, but right now I am crunching 2 regular WUs and only have 1 rush job queued although I have a 0.5 WU buffer set.
----------------------------------------So you're saying that I won't get any other WUs until I actually start crunching the re-issued one or actually complete it? Knowledge is limited. Imagination encircles the world! - Albert Einstein |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
No, it's just backed up. To proof this, just upped the buffer by 0.25 days to 0.5 on the HCC only duo, and promptly got served:
330 World Community Grid 13-03-2013 10:30 [sched_op] CPU work request: 43890.23 seconds; 0.00 devices 331 World Community Grid 13-03-2013 10:31 Scheduler request completed: got 3 new tasks 332 World Community Grid 13-03-2013 10:31 [sched_op] Server version 701 333 World Community Grid 13-03-2013 10:31 Project requested delay of 11 seconds 334 World Community Grid 13-03-2013 10:31 [sched_op] estimated total CPU task duration: 49499 seconds 335 World Community Grid 13-03-2013 10:31 [sched_op] Deferring communication for 11 sec 336 World Community Grid 13-03-2013 10:31 [sched_op] Reason: requested by project 337 World Community Grid 13-03-2013 10:31 Started download of X0900124081229201012160905_X0900124081229201012160905.zip 338 World Community Grid 13-03-2013 10:31 Started download of X0900124081219201012160905_X0900124081219201012160905.zip 339 World Community Grid 13-03-2013 10:31 Finished download of X0900124081229201012160905_X0900124081229201012160905.zip 340 World Community Grid 13-03-2013 10:31 Finished download of X0900124081219201012160905_X0900124081219201012160905.zip 341 World Community Grid 13-03-2013 10:31 Started download of X0900124081225201012160905_X0900124081225201012160905.zip 342 World Community Grid 13-03-2013 10:31 Finished download of X0900124081225201012160905_X0900124081225201012160905.zip Had 4, now 7. If worried of running dry on HCC, up the buffer to be able to bridge these bottlenecks. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Backed up, as in that could include a load of GPU wingman holding the slots until they too get assigned. The price I guess of there only being one feeder for HCC. If that were different, the techs could pitch in.
|
||
|
|
CandymanWCG
Senior Cruncher Romania Joined: Dec 20, 2010 Post Count: 421 Status: Offline Project Badges:
|
Thanks again, SekeRob.
----------------------------------------The reason why I have a 0.5 buffer is precisely to try and avoid these bottlenecks. Until yesterday I was running on 0 buffer because I don't keep my laptop running 24/7 and I almost never switch it on during the weekend so I was afraid of not wasting some good WUs by not returning them on time (and I actually almost got that this Monday with one WU). In the meantime I have just delivered one of the regular WUs that I had running and I got another one lined up, so I guess it worked and I didn't run out of work, like it happened to me (again) on Monday, when I wasted about an hour worth of a core's time. Anyway, I guess that the issue is not really an issue anymore. I will just mark this as "resolved". Cheers! Knowledge is limited. Imagination encircles the world! - Albert Einstein |
||
|
|
|