Message boards : Number crunching : Minirosetta 3.52
Previous · 1 . . . 4 · 5 · 6 · 7
Author | Message |
---|---|
mikey Send message Joined: 5 Jan 06 Posts: 1895 Credit: 9,217,610 RAC: 1,154 |
Sandman192 wrote
Science waits for no man and sometimes needs the answers to the current question to ask the next question, so if you and everyone else wants work everyday then the deadlines must be short enough to allow the different Teams to get the results back and setup the next run of tasks. Sure they could send out a bazillion tasks but WHY when a few hundred thousand will do the job just fine, the answers they are looking for come back well within the zone of acceptance from a few hundred thousand tasks so sending out a bazillion tasks just so you can have a 10 day cache setting is just wasted time and money on your part...and that would be criminal to the Science!! GpuGrid is similar, it uses gpu's only, the next task is based on the last tasks result so they have a 2 at a time max task quota per pc. If that doesn't fit into the way you want to run your pc's then don't run GpuGrid. Have you tried World Community Grid? They have Covid-19 tasks, they call them Open Pandemic tasks, and right now they have plenty for you to have a 10 day cache of tasks on as many pc's as you have. The Project TN-Grid at http://gene.disi.unitn.it/test/ is also doing Covid-19 tasks and they have over 6,000 tasks ready to send out right this minute. The invitation code is right there on their home page in case you've never run it before. |
Sandman192 Send message Joined: 22 Sep 07 Posts: 16 Credit: 2,018,819 RAC: 0 |
It is not 1000%. I don't know where you'r looking at? But under "Resource Share" it's at 1000% for Rosseta. Still doesn't explain: 10/13/2020 11:29:35 PM | Rosetta@home | Tasks won't finish in time: BOINC runs 99.4% of the time; computation is enabled 97.3% of that That has nothing to do with additional days or Resource Share" settings. Did I mention this is happing on 2 computers. Why should I set "Store at least 0.00 days and additional 0.01 days or work"? For one thing the additional days of work is irrelevant since it does not stop Rosseta to not get any work at all. And that's what is just did. Never had this problem before. It just stop getting Rosetta work months ago. And I can't get a second computer to get any Roessta work either. Well it got 1 WU then never got any more after that. |
Sandman192 Send message Joined: 22 Sep 07 Posts: 16 Credit: 2,018,819 RAC: 0 |
You do not want 10 days of work for Rosetta@home. Well, I used BOINC on a 3rd computer with nothing on it. I tried the 10 days or work for Rosetta work for 10 days and then stoped it from getting any more and it finished all of that work that it got. I got at least 15 WUs on a 4 core with 10 days of work set. Task deadlines are always 3 days from when they are delivered. If you ask for 10 days’ worth of work, you might get it – but stand no chance of completing more than 30% of it. All that will do is delay the analysis of the results of the other 70%, as researchers have to wait at least an extra 3 days for the server to resend the tasks to hosts that will actually do the work on time. People, that's not how it works. |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1731 Credit: 18,496,701 RAC: 20,605 |
Well, I used BOINC on a 3rd computer with nothing on it. I tried the 10 days or work for Rosetta work for 10 days and then stoped it from getting any more and it finished all of that work that it got. I got at least 15 WUs on a 4 core with 10 days of work set.So what??? The deadlines are 3 days, so once BOINC works out how much work you want, and how much you can do it then won't let you get more than 3 days worth of work. If you then run multiple other projects, then the amount of Rosetta work you get will be limited as a result of your resource share & your cache settings. You posted here asking why you hadn't got more work with your systems at that time, i told you. If you had this new computer connected to the all the same number of projects with the same settings, and then just let it be, eventually it wouldn't have got any more Rosetta work either, just like happened with your present systems. If you want work, on a fairly regular basis, change your settings as i suggested. If not, then why do you continue posting here with things that aren't relevant? Grant Darwin NT |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1731 Credit: 18,496,701 RAC: 20,605 |
Look very, very, very carefully at your Resource share number in Rosetta- there is no % sign there. That's because, as i said before, it's not a percentage. It's a ratio.It is not 1000%.I don't know where you'r looking at? But under "Resource Share" it's at 1000% for Rosseta. For one thing the additional days of work is irrelevant since it does not stop Rosseta to not get any work at all. And that's what is just did. Never had this problem before.Because of the settings you have chosen that you claim have no effect. The more projects you do & the larger the cache you use the longer it take for your Resource share settings to be honoured. It's taken all this time to get to the point where it doesn't need any more work from Rosetta to meet your Resource share settings, and it probably needs to clear much of the work it presently has from other projects- so they don't run in to deadline issues -before it will be able to do any more Rosetta work. Eventually you will get more Rosetta work, but it could be a long time off still- all because of the settings you have chosen to use. You can choose to fix you settings to something more reasonable- ie no cache & higher share for Rosetta (by lowering the values you've selected for the other projects). Or you can just continue to complain about something that has occurred because you have set it up to do so (regardless of whether or not you believe it). Grant Darwin NT |
Sid Celery Send message Joined: 11 Feb 08 Posts: 2145 Credit: 41,560,787 RAC: 9,320 |
You do not want 10 days of work for Rosetta@home. LOL |
Message boards :
Number crunching :
Minirosetta 3.52
©2024 University of Washington
https://www.bakerlab.org