Estimated time to completion not working correctly

Message boards : Number crunching : Estimated time to completion not working correctly

To post messages, you must log in.

AuthorMessage
Profile Angus

Send message
Joined: 17 Sep 05
Posts: 412
Credit: 321,053
RAC: 0
Message 11072 - Posted: 21 Feb 2006, 6:49:19 UTC

Now that the RALPH beta project has finished (?) and the new client is in place on the live Rosetta project, the crunch time per WU on the Work tab in BOINC Manager should accurately reflect the preference set in the Rosetta preference screen.

I set my preference to 2 hours (the lowest possible), and attached to the live Rosetta project.

BOINC downloaded a handful of WUs, and every one had 5:40 as the time to completion, and it was nowhere near enough WUs to fill the 10 day queue.

I did a <RESET> , <UPDATE> to make sure the prefences were loaded, and still got 5:40 on the WUs.

What's the deal? If the fixed crunch time is 2 hours, why don't I get the 120 WUs it should take to fill my 10 day queue?

Somethings broken.
Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)
ID: 11072 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Scribe
Avatar

Send message
Joined: 2 Nov 05
Posts: 284
Credit: 157,359
RAC: 0
Message 11073 - Posted: 21 Feb 2006, 6:55:05 UTC

It takes time before the figure adjusts itself, you have to process a good number before it gets there....it is the fault of the RDCF....
ID: 11073 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile David E K
Volunteer moderator
Project administrator
Project developer
Project scientist

Send message
Joined: 1 Jul 05
Posts: 1018
Credit: 4,334,829
RAC: 0
Message 11075 - Posted: 21 Feb 2006, 7:07:10 UTC

This is a boinc client issue. We do not have a way to modify the rsc_fpops_est (the value used by the client to estimate the time to completion). As Scribe mentioned below, the correction factor should eventually make the time to completion more accurate.

RALPH IS NOT FINISHED. IT IS AN ONGOING PROJECT FOR TESTING.
ID: 11075 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Angus

Send message
Joined: 17 Sep 05
Posts: 412
Credit: 321,053
RAC: 0
Message 11077 - Posted: 21 Feb 2006, 7:11:42 UTC

If the time to completion is user-defined, why is the client doing an estimate?
Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)
ID: 11077 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Angus

Send message
Joined: 17 Sep 05
Posts: 412
Credit: 321,053
RAC: 0
Message 11078 - Posted: 21 Feb 2006, 7:13:05 UTC - in response to Message 11073.  

It takes time before the figure adjusts itself, you have to process a good number before it gets there....it is the fault of the RDCF....

RDC should not even be considered in this case. The time to completion is user defined. The WU unit runs for as long as the user has set in the preferences.

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)
ID: 11078 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile David E K
Volunteer moderator
Project administrator
Project developer
Project scientist

Send message
Joined: 1 Jul 05
Posts: 1018
Credit: 4,334,829
RAC: 0
Message 11080 - Posted: 21 Feb 2006, 7:21:56 UTC
Last modified: 21 Feb 2006, 7:24:58 UTC

This feature/preference is unique to our project and the boinc client is not yet designed to use this user defined preference for the time to completion. Maybe the boinc developers can add this into the next client version. The DCF is, thus, still relevant for the time to competion.
ID: 11080 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Moderator9
Volunteer moderator

Send message
Joined: 22 Jan 06
Posts: 1014
Credit: 0
RAC: 0
Message 11108 - Posted: 21 Feb 2006, 13:03:15 UTC - in response to Message 11080.  

This feature/preference is unique to our project and the boinc client is not yet designed to use this user defined preference for the time to completion. Maybe the boinc developers can add this into the next client version. The DCF is, thus, still relevant for the time to competion.



Also remember for the shorter run time settings, the WU will always complete at least one model. In some cases that one model may take a lot longer than 2 hours. During the time it takes to run the first model, the WU will only show 1% complete. It might look like the WU is hung, but if you check the screen saver or image display, you will see the steps running. On some WU a model only take about 35,000 steps and 5 min. others may take over 500,000 steps and take 3-4 hours to do a single model. So if you have your time set to 2 hours, the WU will run way over that to do the first model. This means that you could have more work on your system than you think you have.

Moderator9
ROSETTA@home FAQ
Moderator Contact
ID: 11108 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Angus

Send message
Joined: 17 Sep 05
Posts: 412
Credit: 321,053
RAC: 0
Message 11119 - Posted: 21 Feb 2006, 15:59:22 UTC

I thought that the estimated time to completion was driven by something in the WU. Different WUs from different projects get different estimated times.
That sounds like a project-specific thing, not a BOINC Manager issue. The rosetta client should be altering the estimated time to match the user preference.
Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)
ID: 11119 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Moderator9
Volunteer moderator

Send message
Joined: 22 Jan 06
Posts: 1014
Credit: 0
RAC: 0
Message 11121 - Posted: 21 Feb 2006, 16:18:50 UTC - in response to Message 11119.  

I thought that the estimated time to completion was driven by something in the WU. Different WUs from different projects get different estimated times.
That sounds like a project-specific thing, not a BOINC Manager issue. The rosetta client should be altering the estimated time to match the user preference.




It is a combination of things. How long you set the time setting tells the application how long you want the WUs to run.

But it only does so based on the length of a model run for that particular flavor of WU. The shortest run time would be the length of 1 model generation. So if you have a flavor of WU that can do a model in say 5 min. You can set the time to 2 hours and the WU will complete in about that amount of time. But if the particular type of WU would take say 4 hours to do 1 model, then that WU will run for at least 4 hours no matter what the time setting is.


Moderator9
ROSETTA@home FAQ
Moderator Contact
ID: 11121 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Scribe
Avatar

Send message
Joined: 2 Nov 05
Posts: 284
Credit: 157,359
RAC: 0
Message 11122 - Posted: 21 Feb 2006, 16:19:40 UTC - in response to Message 11119.  
Last modified: 21 Feb 2006, 16:20:01 UTC

I thought that the estimated time to completion was driven by something in the WU. Different WUs from different projects get different estimated times.
That sounds like a project-specific thing, not a BOINC Manager issue. The rosetta client should be altering the estimated time to match the user preference.


Still not reading the answers are we??? Why not read the ones given to you?

This feature/preference is unique to our project and the boinc client is not yet designed to use this user defined preference for the time to completion. Maybe the boinc developers can add this into the next client version. The DCF is, thus, still relevant for the time to competion.


ID: 11122 · Rating: -1 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile David E K
Volunteer moderator
Project administrator
Project developer
Project scientist

Send message
Joined: 1 Jul 05
Posts: 1018
Credit: 4,334,829
RAC: 0
Message 11141 - Posted: 21 Feb 2006, 19:28:05 UTC - in response to Message 11119.  

I thought that the estimated time to completion was driven by something in the WU. Different WUs from different projects get different estimated times.
That sounds like a project-specific thing, not a BOINC Manager issue. The rosetta client should be altering the estimated time to match the user preference.


We are only able to interact with the boinc client/manager via the boinc api with our application. The api does not allow us yet to modify the time to competion value during a run so we cannot change it when a user does so. You are correct, the estimated time to completion is driven by something in the WU (rsc_fpops_est) but it is set when the work unit is generated and once it is downloaded onto your client we cannot change the value since the boinc api does not let us yet.

See http://boinc.berkeley.edu/api.php. This is what we have to work with.
ID: 11141 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Estimated time to completion not working correctly



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