| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: OpenPandemics - COVID-19 Project
Thread: Task Run Time |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 7
|
|
| Author |
|
|
highwind
Advanced Cruncher Joined: Aug 22, 2009 Post Count: 56 Status: Offline Project Badges: 
|
I am currently running COVID19 on my 3900X overclocking it to 4.25Ghz,
I noticed that most tasks are finished within 105 mins but there are always a few tasks running north of 140 mins. This is also being observed on my laptop 7700HQ where most tasks are done within 3 hours and some are north of 4 hours. This is quite inconsistent with other project like MCM where all tasks are finished within 90 mins on my 3900X. Anyone seeing the see same thing? |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7846 Status: Offline Project Badges: 
|
Yes, the work units are not all the same length. Some dockings are more complicated.
----------------------------------------Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
yoerik
Senior Cruncher Canada Joined: Mar 24, 2020 Post Count: 413 Status: Offline Project Badges:
|
yep. On android (where runtimes are far longer) - I've noticed some run twice as fast as others. Some take 8 - others take well over 12 on my phone - on my old phone and tablet - some run in 15, some run 20+. You should see the differences if you look at the result log after they're complete - of course, once the stats update is done.
---------------------------------------- |
||
|
|
Speedy51
Veteran Cruncher New Zealand Joined: Nov 4, 2005 Post Count: 1326 Status: Offline Project Badges: 
|
I did an experiment with hyperthreading on verse off using same CPU running Windows. I found with out HT a task would take 1.00 hour with HT on a job would take about 1.70 hours.
---------------------------------------- [Edit 1 times, last edit by Speedy51 at Jun 20, 2020 5:35:13 AM] |
||
|
|
DrMason
Senior Cruncher Joined: Mar 16, 2007 Post Count: 153 Status: Offline Project Badges: 
|
I also did some tests on hyperthreading, but with amd epyc processors running linux mint. There, after I disabled hyperthreading, I found that tasks finished just barely more than twice as fast as with hyperthreading on. So, on some architectures and operating systems, it appears that hyperthreading (while great for getting those badges) actually produces less overall work. Maybe the fact that it's using less power and producing less heat means that the processors are able to boost a little better? I'm not sure, so I'd encourage people to do some experiments and see what works best for them.
---------------------------------------- |
||
|
|
Bobzilla
Cruncher Joined: May 31, 2020 Post Count: 31 Status: Offline Project Badges:
|
Don't know if this helps or not but I also have a 3900x with 64gb ram. I keep it stock mainly because Im afraid to overclock and mess something up. My run times are 1hr 20min to 1hr 57min I guess it varies on what it's working on. It runs 24 at a time, and I have selected to do 1 day work and download 2nd day work so I always have plenty to do. Hope this helps, sorry if it doesn't.
----------------------------------------Quick edit: Just looked at actual times. they range from 1 hr 2 mins to 1 hr 52 mins and that is for 24 ready to report. i am starting to think that it's not JUST the CPU alone that lowers the time, but maybe the available ram for each thread to compile the task that they work on? Remember: With beer, all things are possible. Some of which, probably should not be tried. [Edit 1 times, last edit by Bobzilla at Jun 1, 2020 6:04:57 AM] |
|||
|
||||
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1403 Status: Offline Project Badges: 
|
Since running 4 tasks on my 8-threaded Android, the total runtimes are rather consistent between 5 and 6 hours.
The inside number of jobs and dockings varies more: ZINC001362836299-ACR3.38_RX1--6w9c_001_tyr264-HO--CYS111_wcgsplit3.dpf(Job #0)... 10 # Dockings Total sec. Job #0 seems to be most of the time a remnant from a previous task done by someone else. |
||
|
|
|