Message boards : Number crunching : "Completed, too late to validate" - result of too short estimated runtime?
Author | Message |
---|---|
Chris Send message Joined: 12 Apr 06 Posts: 6 Credit: 13,598,060 RAC: 0 |
Hi all, For about a week now, I'm contributing next to nothing, maybe because: - estimated runtime for my rig is about 5 hrs per task - real runtime for most WUs is between 20 and 24 hrs - deadline for new WUs is very short As a result, majority of WUs are either aborted or not validated, because they finish too late. I've set my job cache to 0.8 + 0.1, two days ago, but today morning it's still the same, 24hrs of crunching for nothing, too late. There are not started WUs in queue with deadline today, which I know I have to abort, or it will be waste of cpu. I'm using two PCs with Win10, one is 6800K, the other 9700K, both with similar issues. What I could possibly do to make sure those WUs are processed and validated? Chris |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1677 Credit: 17,751,802 RAC: 22,980 |
What I could possibly do to make sure those WUs are processed and validated?It all depends on what values your other settings have. For starters, go to your Account, Rosetta@home preferences, make sure "Target CPU run time" is "not selected" and save the changes. Hit update on the BOINC Manager for changes to take effect. That will mean the Target CPU Runtime for any new Tasks to start will be 8 hours (the default). Already started Tasks will continue to run to the previous setting. I've set my job cache to 0.8 + 0.1, two days ago,If you're running more than one project, that is still too big. With your computer(s) hidden it's not possible to see what else is going on, but i'd suggest the following settings, particularly the much smaller cache setting Computing Usage limits Use at most 100% of the CPUs Use at most 100% of CPU time When to suspend Suspend when computer is on battery (not selected) Suspend when computer is in use (not selected) Suspend GPU computing when computer is in use (not selected) 'In use' means mouse/keyboard input in last 3 minutes Suspend when no mouse/keyboard input in last --- minutes Suspend when non-BOINC CPU usage is above --- % Compute only between --- Other Store at least 0.3 days of work Store up to an additional 0.02 days of work Switch between tasks every 60 minutes Request tasks to checkpoint at most every 60 seconds Disk Use no more than 20 GB Leave at least 2 GB free Use no more than 60 % of total Memory When computer is in use, use at most 95 % When computer is not in use, use at most 95 % Leave non-GPU tasks in memory while suspended (not selected) Page/swap file: use at most 75 %It will still take some time for the Estimated completion times to match the Target CPU Runtime, so some Taks may still time out, but it will help stop you from getting even more work that you won't be able to finish. And it will allow the BOINC Manager to eventually honour your Resource share settings once the Estimated completion times get closer to reality. Use at most 100% of the CPUs may cause memory problems with different Tasks that require more RAM, but with your system(s) hidden there's no way of knowing. Grant Darwin NT |
Chris Send message Joined: 12 Apr 06 Posts: 6 Credit: 13,598,060 RAC: 0 |
Many thanks Grant, My computers should be visible now. I'm running only R@H these days after SETI closed, but anyway I followed your suggestion about job cache. The older PC is running 100%CPU/Time, but it has liquid cooling, the newer one runs 6 out of 8 cores, since it's getting too hot and loud. Best regards from literally other side of the globe ;) Chris |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1677 Credit: 17,751,802 RAC: 22,980 |
I'm running only R@H these days after SETI closed, but anyway I followed your suggestion about job cache.It'd be worth making sure the Target CPU Runtime in the Rosetta project preferences is set to 8 hours (or better yet "not selected" so it can change according to the project's requirements). The older PC is running 100%CPU/Time, but it has liquid cooling, the newer one runs 6 out of 8 cores, since it's getting too hot and loud.Good choice- if things get hot or noisy, limiting cores is better than reducing "Use at most xx % of CPU time" from 100%. Both systems have plenty of RAM for the number of cores/threads so you won't run in to problems when the next group of large RAM requirement Tasks come through (roughly 1.3GB RAM per Task). But i did notice an issue with the i7-6800K system. Run time 1 days 1 hours 57 min 50 sec CPU time 1 days 0 hours 2 min 34 secWith 24hours of Runtime, i'd expect the difference between CPU time & Runtime to be around 10min or so, not 2hrs. Is this the system you use during the day? Do you make use of any CPU intensive programmes? Because a difference between CPU time & Runtime that large shows the CPU is doing lots of work other than crunching BOINC work. If you're making lots of CPU heavy use of the system, no problem. If not, you might want to use Task Manager (or Process Explorer) to see what is taking up that chunk of CPU time from BOINC work. Grant Darwin NT |
Message boards :
Number crunching :
"Completed, too late to validate" - result of too short estimated runtime?
©2024 University of Washington
https://www.bakerlab.org