| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Smash Childhood Cancer
Thread: Performance issues |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 12
|
|
| Author |
|
|
ngsmith
Cruncher Joined: Jul 21, 2005 Post Count: 48 Status: Offline Project Badges: 
|
I wonder if anyone else is having performance issues on SCC. System is a Dell XPS i7-3770 @ 3.4 GHZ, 12 GB RAM, SSD boot device, 2 GB SATA drive for applications. Recently upgraded to Win10 version 1803.
----------------------------------------BOINC manager set to 100% of the CPUs, 70% of CPU time. Until last few days, Task Manger has been running at 70% load. Anyways, SCC tasks have been running very slow - predicted 3 hr jobs are 7 days remaining after 8 hrs run. I've aborted tasks, and the resends are completed in 1-2 hrs. CPU load is running at less than 30%. Aborting long running tasks get back to 70% CPU load. I'd like to get to 2 years, but at this rate, it ain't gonna happen.  |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7845 Status: Recently Active Project Badges: 
|
I wonder if anyone else is having performance issues on SCC. System is a Dell XPS i7-3770 @ 3.4 GHZ, 12 GB RAM, SSD boot device, 2 GB SATA drive for applications. Recently upgraded to Win10 version 1803. BOINC manager set to 100% of the CPUs, 70% of CPU time. Until last few days, Task Manger has been running at 70% load. Anyways, SCC tasks have been running very slow - predicted 3 hr jobs are 7 days remaining after 8 hrs run. I've aborted tasks, and the resends are completed in 1-2 hrs. CPU load is running at less than 30%. Aborting long running tasks get back to 70% CPU load. I'd like to get to 2 years, but at this rate, it ain't gonna happen. I presume you have a desktop. rather than try to run all the cores at 70%, why not just use 6 cores at 100%. You should not have to worry about overheating I would think. I have a feeling that the resume on the on-again off-again of the way BOINC handles the throttling is causing one or more of the threads to hang. I don't run Win 10, so this is a guess. All the tasks I have seen of this project have run in 2 to 2 1/2 hour range on older equipment than you have. One other thing comes to mind. If you have the long running task, which I believe is hung, try rebooting to see if that frees up the task to a more normal time. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
sunk818
Advanced Cruncher Joined: May 10, 2018 Post Count: 66 Status: Offline Project Badges: 
|
ngsmith - If you're running i7-3770 you should be getting 8k-9k RAC.
Maybe restart your computer and see what happens. Also, check to see if any other background processes are running like Windows Defender, anti-virus, or any other processes hogging up resources. If you're going to crunch SCC though, you should run it on Linux. I'm learning how to install VirtualBox and Ubuntu 16 just to get the crunching advantage on SCC. |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges: 
|
His post was back in May and from looking at his sig, it look like he has given up the ghost.
---------------------------------------- |
||
|
|
True54Blue
Advanced Cruncher Joined: Nov 17, 2004 Post Count: 97 Status: Offline Project Badges:
|
ngsmith did get to 2 years on SCC judging by his badge. Kudos to that.
---------------------------------------- |
||
|
|
TPCBF
Master Cruncher USA Joined: Jan 2, 2011 Post Count: 2173 Status: Offline Project Badges:
|
Anyways, SCC tasks have been running very slow - predicted 3 hr jobs are 7 days remaining after 8 hrs run. I've aborted tasks, and the resends are completed in 1-2 hrs. CPU load is running at less than 30%. Aborting long running tasks get back to 70% CPU load. Could be related to the issue that I see on various PCs here as well, unfortunately, a couple of them at least without direct access to them.WUs will run up to the very high 90% range and then just "sit" until they get aborted by me (if I get a chance) or they simply time out and result in a "no reply". Common across all the instances where I noticed this is that they show in the properties for the task in the BOINC manager for "CPU Time" and "CPU Time since last checkpoint" just a few dashes. On one remote laptop (i7, 8 threads, 8GB of RAM), that sits 95% of the day idle on a desk, I recently noticed 7 of such WUs clogging up the queues for days, with just one slot slogging along returning WUs just as normal. And I always wondered why there are at times noticable drops in the daily returns. I suspect that this happens on even more machines than the 5 or 6 where I have observed this so far. I think I mentioned this earlier this year already, but there was no reasonable reply as to the source of the issue... :-( Ralf |
||
|
|
l_mckeon
Senior Cruncher Joined: Oct 20, 2007 Post Count: 439 Status: Offline Project Badges: 
|
Have you tried turning it off then on again (not joking)?
Sometime these stuck tasks will then restart from the last saved checkpoint. |
||
|
|
Pete Broad
Senior Cruncher Wales Joined: Jan 3, 2007 Post Count: 169 Status: Offline Project Badges:
|
Have you tried turning it off then on again (not joking)? Sometime these stuck tasks will then restart from the last saved checkpoint. See this on one of my win 10 machines. Simple solution, exit boinc, restart boinc, works for me. Pete  [Edit 1 times, last edit by Pete Broad at Dec 7, 2018 8:33:30 PM] |
||
|
|
TPCBF
Master Cruncher USA Joined: Jan 2, 2011 Post Count: 2173 Status: Offline Project Badges:
|
Have you tried turning it off then on again (not joking)? The point that you missed is that there is no "last saved checkpoint" on those WUs!Sometime these stuck tasks will then restart from the last saved checkpoint. Shutting down the BOINC agents and restarting will start them from scratch, most of the time, just to repeat the process. No checkpoint and no logged CPU time, just wasting slots inaccessible for others. If it happens on one of the PCs that I have direct access to, I just abort the WU, which will then turn into a "computation error". On remote machines, without remote access, those just clog the system until their deadline expires... Ralf |
||
|
|
TPCBF
Master Cruncher USA Joined: Jan 2, 2011 Post Count: 2173 Status: Offline Project Badges:
|
Have you tried turning it off then on again (not joking)? Sometime these stuck tasks will then restart from the last saved checkpoint. See this on one of my win 10 machines. Simple solution, exit boinc, restart boinc, works for me. Pete As mentioned, the real problem are those hosts that I don't constantly monitor, the just stuff up the system. And other WUs, depending on how many slots/threads are still workable, will work just fine, sometime hundreds of them before the "bad one" will expire and kicked out by BOINC when the deadline passes. And that can be for 10 days. One one recent system (i7, 8GB RAM, barely in use most of the day), there were 7 hung WUs and just one that was crunching normal. Pretty frustrating... Ralf |
||
|
|
|