Problems and Technical Issues with Rosetta@home

Message boards : Number crunching : Problems and Technical Issues with Rosetta@home

To post messages, you must log in.

Previous · 1 . . . 46 · 47 · 48 · 49 · 50 · 51 · 52 . . . 310 · Next

AuthorMessage
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 95525 - Posted: 29 Apr 2020, 3:14:50 UTC - in response to Message 95523.  

There has been talk, but not yet action, about a new class of workunits that can use up to 4 gigabytes of memory each, rather than the usual up to 2 gigabytes. This is intended to allow work on larger proteins, which will probably also require larger target times,


Actually, Admin said the models for those particular large proteins should only run about a hour, and typically consumed 1.8GB. I had presumed they would take longer as well, so I apologize for contributing to the mistaken information. It all depends on the type of study they perform on the protein. In this case, they are planning to do "comparative modeling" on the large proteins.
Rosetta Moderator: Mod.Sense
ID: 95525 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile robertmiles

Send message
Joined: 16 Jun 08
Posts: 1234
Credit: 14,338,560
RAC: 1,227
Message 95526 - Posted: 29 Apr 2020, 3:30:09 UTC - in response to Message 95525.  
Last modified: 29 Apr 2020, 3:31:53 UTC

There has been talk, but not yet action, about a new class of workunits that can use up to 4 gigabytes of memory each, rather than the usual up to 2 gigabytes. This is intended to allow work on larger proteins, which will probably also require larger target times,


Actually, Admin said the models for those particular large proteins should only run about a hour, and typically consumed 1.8GB. I had presumed they would take longer as well, so I apologize for contributing to the mistaken information. It all depends on the type of study they perform on the protein. In this case, they are planning to do "comparative modeling" on the large proteins.

Correction - models needing up to 4 gigabytes are talked about for SOME BOINC project, but I'm not sure if it is Rosetta@home.
ID: 95526 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 95527 - Posted: 29 Apr 2020, 3:37:41 UTC - in response to Message 95526.  
Last modified: 29 Apr 2020, 3:39:00 UTC

R@h has talked about "4GB tasks"... where they are telling the BOINC Manager "kill this task if it should ever try to go larger than 4GB of memory, something is wrong with it if it is going that large (memory bound)". The actual observed footprint though is typically 1.8GB.
Rosetta Moderator: Mod.Sense
ID: 95527 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Michael E.@ team Carl Sagan

Send message
Joined: 5 Apr 08
Posts: 16
Credit: 1,947,553
RAC: 128
Message 95528 - Posted: 29 Apr 2020, 4:31:36 UTC - in response to Message 95527.  

Many thanks to Mod.Sense, robertmiles, CIA and (previously) Grant for the clear explanations.

So it seems R@h users should stick to the default task size, which is 8 hours. For older systems or those not used 24x7, choose shorter length tasks.

OK about the system learning the Remaining time after about 12 tasks (good!). I do think limiting the number of work units for new hosts that have not run a dozen tasks would help.

Also, I ran into a strange situation where a 24-hour work unit reached zero Remaining time but kept processing for 10 extra hours. Work unit was 1043928617 and task 1043928617.
I had to restart my PC and that task reset to 10 hours remaining so I aborted it as it was out of time.

If you want a beta tester for the large R@h tasks, let me know.
ID: 95528 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 95529 - Posted: 29 Apr 2020, 5:14:10 UTC - in response to Message 95528.  

If you want a beta tester for the large R@h tasks, let me know.


Folks that enjoy this sorta thing can join Ralph@home. This project is run by the same folks as R@h. Ralph sorta short for "Rosetta Alpha". Works just like any other BOINC project. It is where new application versions, server configurations, and work unit types are tested. They only send work when they need testers though, so get connected and let the BOINC scheduler keep asking for work periodically, and eventually you get some.
Rosetta Moderator: Mod.Sense
ID: 95529 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Daedalus

Send message
Joined: 1 Aug 08
Posts: 39
Credit: 10,107,661
RAC: 34
Message 95558 - Posted: 29 Apr 2020, 17:34:56 UTC

I get an awful lot of tasks that complete very slowly. I set my runtime goal to 4 hours. Most rotten tasks advance at exactly 6.480 % an hour. I kill them mercilessly. But i have to babysit rosetta and check all my tasks one by one as soon as they start crunching. This is a waste of my time and my electricity. Should i dump my whole queue ?

P.S. And previously i was getting a lot of tasks with a completion time of 9 hours. But you could detect them and dump them before they started so less effort was needed.
ID: 95558 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Millenium

Send message
Joined: 20 Sep 05
Posts: 68
Credit: 184,283
RAC: 0
Message 95559 - Posted: 29 Apr 2020, 18:28:09 UTC

Or maybe they are just fine and it takes more than 4 hours to complete. The default runtime is 8 hours after all. What happens if you let them run? Do they keep going for days? Do they eventually end?
ID: 95559 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Daedalus

Send message
Joined: 1 Aug 08
Posts: 39
Credit: 10,107,661
RAC: 34
Message 95562 - Posted: 29 Apr 2020, 19:56:26 UTC

Ok, i may have an optimistic explanation of this: The completion percentages shown in the BOINC manager might be wildly innacurate.

I had "very slow" WU's and "less slow" WU's. I killed a lot of very slow ones and let the less slow run their course. Surprise: the less slow tasks, announced to take 6 hours or more have stopped at around 4 hours and half as promised. They were at 85 -90 % completion and they suddently jumped to 100% With the correct run time.

Let's hope it will be the same with the "very slow" ones. I will only know tomorrow.
ID: 95562 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile robertmiles

Send message
Joined: 16 Jun 08
Posts: 1234
Credit: 14,338,560
RAC: 1,227
Message 95566 - Posted: 29 Apr 2020, 20:41:58 UTC - in response to Message 95563.  

Daedalus,

The estimated completion times usually ARE wildly inaccurate for about the next dozen tasks after any of the following events occur:

1. You adjust your target time on the Rosetta@home server.

2. You start using a new version of a Rosetta@home application.

3. Any other actions making a big difference in how long the tasks run. For example, having Folding@home and Rosetta@home running at the same time without doing enough to eliminate their competition for CPU time.
ID: 95566 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Tom M

Send message
Joined: 20 Jun 17
Posts: 94
Credit: 16,625,100
RAC: 51,990
Message 95572 - Posted: 29 Apr 2020, 23:30:00 UTC - in response to Message 95562.  

Ok, i may have an optimistic explanation of this: The completion percentages shown in the BOINC manager might be wildly innacurate.

I had "very slow" WU's and "less slow" WU's. I killed a lot of very slow ones and let the less slow run their course. Surprise: the less slow tasks, announced to take 6 hours or more have stopped at around 4 hours and half as promised. They were at 85 -90 % completion and they suddently jumped to 100% With the correct run time.

Let's hope it will be the same with the "very slow" ones. I will only know tomorrow.


Sounds reasonable. Just wait for it :)

Tom M
Help, my tagline is missing..... Help, my tagline is......... Help, m........ Hel.....
ID: 95572 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Daedalus

Send message
Joined: 1 Aug 08
Posts: 39
Credit: 10,107,661
RAC: 34
Message 95588 - Posted: 30 Apr 2020, 9:34:38 UTC

Yes, save a few tasks cancelled by the server, all my tasks computed normally. So i wil ignore the progression as reported by the BOINC manager. :)
ID: 95588 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Millenium

Send message
Joined: 20 Sep 05
Posts: 68
Credit: 184,283
RAC: 0
Message 95596 - Posted: 30 Apr 2020, 11:18:47 UTC

Yup, as long as the WUs finish it's fine. Sure, if you see one running for a day and more then maybe that WU has a problem.
ID: 95596 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 2145
Credit: 41,555,266
RAC: 8,961
Message 95612 - Posted: 30 Apr 2020, 15:17:03 UTC - in response to Message 95517.  

Shall do. Step 1 - complete the tasks I have now. Then download more if successful.
Thanks.

It seems like you're not running tasks 1215hrs every day. Or if you are, you maybe have kind of setting to suspend work while the computer is in operation?
It's not clear why nothing is returned within 3 days when you only have 8hr tasks if your machine is on 1215hrs a day.
ID: 95612 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
James W

Send message
Joined: 25 Nov 12
Posts: 130
Credit: 1,766,254
RAC: 0
Message 95755 - Posted: 2 May 2020, 3:44:08 UTC

Today I started receiving the following message in BOINC Manager (v 7.16.5) Event Log, as well as in BOINC Notices.
5/1/2020 8:08:57 PM | Rosetta@home | This project is using an old URL. When convenient, remove the project, then add https://boinc.bakerlab.org/rosetta/

Is it really necessary to remove the project to change URL? Doing this will remove all my current and pending tasks and I'd have to reload from square-one. Correct? Another way to fix this issue?
ID: 95755 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Grant (SSSF)

Send message
Joined: 28 Mar 20
Posts: 1729
Credit: 18,490,957
RAC: 20,862
Message 95756 - Posted: 2 May 2020, 3:51:01 UTC - in response to Message 95755.  

Today I started receiving the following message in BOINC Manager (v 7.16.5) Event Log, as well as in BOINC Notices.
5/1/2020 8:08:57 PM | Rosetta@home | This project is using an old URL. When convenient, remove the project, then add https://boinc.bakerlab.org/rosetta/
Is it really necessary to remove the project to change URL? Doing this will remove all my current and pending tasks and I'd have to reload from square-one. Correct? Another way to fix this issue?
Set No New Tasks.
When all Tasks have been completed & returned, then Remove & re-attach to the project.
When re-attaching to the project, select the "Existing user option." (or whatever it is actually called).
Grant
Darwin NT
ID: 95756 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile robertmiles

Send message
Joined: 16 Jun 08
Posts: 1234
Credit: 14,338,560
RAC: 1,227
Message 95757 - Posted: 2 May 2020, 3:59:04 UTC - in response to Message 95755.  

Today I started receiving the following message in BOINC Manager (v 7.16.5) Event Log, as well as in BOINC Notices.
5/1/2020 8:08:57 PM | Rosetta@home | This project is using an old URL. When convenient, remove the project, then add https://boinc.bakerlab.org/rosetta/

Is it really necessary to remove the project to change URL? Doing this will remove all my current and pending tasks and I'd have to reload from square-one. Correct? Another way to fix this issue?

You can set No new tasks, wait for all current tasks to finish, return those, THEN follow the above instructions before turning off No new tasks.

I've done this on other BOINC projects, causing no problems other than a few hours with no tasks for the affected projects running.

It MIGHT be a good way to delete a few hundred megabytes of obsolete R@h files.
ID: 95757 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 2145
Credit: 41,555,266
RAC: 8,961
Message 95801 - Posted: 2 May 2020, 13:33:35 UTC - in response to Message 95756.  

Today I started receiving the following message in BOINC Manager (v 7.16.5) Event Log, as well as in BOINC Notices.
5/1/2020 8:08:57 PM | Rosetta@home | This project is using an old URL. When convenient, remove the project, then add https://boinc.bakerlab.org/rosetta/
Is it really necessary to remove the project to change URL? Doing this will remove all my current and pending tasks and I'd have to reload from square-one. Correct? Another way to fix this issue?
Set No New Tasks.
When all Tasks have been completed & returned, then Remove & re-attach to the project.
When re-attaching to the project, select the "Existing user option." (or whatever it is actually called).

So I've realised. Thanks.
Of course, it's also possible to abort all non-running Rosetta tasks to make the process of running down the cache much quicker.
I may do that so removingre-attaching is done at my convenience and not in the middle of the night.
ID: 95801 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile GoldenHat

Send message
Joined: 14 Apr 20
Posts: 3
Credit: 122,663
RAC: 0
Message 95989 - Posted: 4 May 2020, 6:36:32 UTC - in response to Message 95801.  

Thanks, very helpful.
Could you also explain how one cleans out the cache and old files etc?
Thanks.
ID: 95989 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
MarkJ

Send message
Joined: 28 Mar 20
Posts: 72
Credit: 25,238,680
RAC: 0
Message 95992 - Posted: 4 May 2020, 7:00:27 UTC - in response to Message 95989.  

Thanks, very helpful.
Could you also explain how one cleans out the cache and old files etc?
Thanks.

That happens when you Reset or Detach (Remove on the BOINC Manager screen) the project.
BOINC blog
ID: 95992 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
James W

Send message
Joined: 25 Nov 12
Posts: 130
Credit: 1,766,254
RAC: 0
Message 96002 - Posted: 4 May 2020, 10:17:21 UTC - in response to Message 95755.  

Today I started receiving the following message in BOINC Manager (v 7.16.5) Event Log, as well as in BOINC Notices.
5/1/2020 8:08:57 PM | Rosetta@home | This project is using an old URL. When convenient, remove the project, then add https://boinc.bakerlab.org/rosetta/

Is it really necessary to remove the project to change URL? Doing this will remove all my current and pending tasks and I'd have to reload from square-one. Correct? Another way to fix this issue?

As a followup, let me state how this process worked for me. Note that I use BOINCstatsBAM as my account manager. I marked Rosetta project "No new tasks" in my host BOINC manager so I could complete jobs in cache before deleting and replacing project with current URL address. I later noticed a note had been added next to "no new tasks" in Project tab that when all tasks completed the project would be deleted and ready for replacement (I've paraphrased exact wording). Sure enough, after last Rosetta task completed and next time host reported to account manager, Rosetta was taken out of my project list. Next time my host reported to account manager, Rosetta was reinstalled with correct info and I was given a starter set of jobs for cache. I was surprised! Not much I had to do other than be sure host synchronized with acct manager. Note that I had previously updated to BOINC manager v7.16.5.
ID: 96002 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Previous · 1 . . . 46 · 47 · 48 · 49 · 50 · 51 · 52 . . . 310 · Next

Message boards : Number crunching : Problems and Technical Issues with Rosetta@home



©2024 University of Washington
https://www.bakerlab.org