[Request] Progress Estimation

Message boards : Number crunching : [Request] Progress Estimation

To post messages, you must log in.

AuthorMessage
Christoph

Send message
Joined: 10 Dec 05
Posts: 57
Credit: 1,512,386
RAC: 0
Message 36248 - Posted: 7 Feb 2007, 16:35:43 UTC

Hi,

the current progress estimation is rather confusing to new participants and a bit uninformative. What about calculating the progress with the help of the WU duration?
percentage = [cpu time] / [wu length set by the user] * 100%

Wouldn't that be more informative and logical?
ID: 36248 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
FluffyChicken
Avatar

Send message
Joined: 1 Nov 05
Posts: 1260
Credit: 369,635
RAC: 0
Message 36256 - Posted: 7 Feb 2007, 20:00:05 UTC - in response to Message 36248.  

Hi,

the current progress estimation is rather confusing to new participants and a bit uninformative. What about calculating the progress with the help of the WU duration?
percentage = [cpu time] / [wu length set by the user] * 100%

Wouldn't that be more informative and logical?


That would still be approximate since it may stop after 4 hrs on a 6 hr task length and I'm sure you'll get some 101%+ situations. But isn't that what is used already or somthing very similar anyway.



Team mauisun.org
ID: 36256 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Christoph

Send message
Joined: 10 Dec 05
Posts: 57
Credit: 1,512,386
RAC: 0
Message 36282 - Posted: 8 Feb 2007, 12:58:16 UTC - in response to Message 36256.  

I'm sure you'll get some 101%+ situations.

You would have to check if the cpu time is more than the target wu length.
But isn't that what is used already or somthing very similar anyway.

Yes, but it's only updated when a model completes and starts at 1%. It would be more logical if it starts at 0% and is updated all the time.
ID: 36282 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Feet1st
Avatar

Send message
Joined: 30 Dec 05
Posts: 1755
Credit: 4,690,520
RAC: 0
Message 36284 - Posted: 8 Feb 2007, 14:37:23 UTC - in response to Message 36248.  

percentage = [cpu time] / [wu length set by the user] * 100%

Wouldn't that be more informative and logical?


I think that's a great idea! As Fluffy points out, it would have it's own quirks, but... for a new participant, it would take longer to find THOSE then it would to see the time to completion INCREASING, and panic and abort! ...which some people do!

I would simply revise the forumla to avoid reaching oer 100%
percentage = [cpu time] / [ 1.05 * (wu runtime preference)] * 100%

or you could compute it sort of backwards
time remaining = [1.05 - (cpu time / wu runtime pref)] * wu runtime pref

The project team actually has embedded some special meanings in to the completion percentages displayed... it let's them more fully understand where the WU is in the program if you report the full value for % complete... but we haven't needed them for a very very long time. (i.e. no more 1% bug!)
Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/
ID: 36284 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Christoph

Send message
Joined: 10 Dec 05
Posts: 57
Credit: 1,512,386
RAC: 0
Message 36292 - Posted: 8 Feb 2007, 15:53:28 UTC - in response to Message 36284.  

The project team actually has embedded some special meanings in to the completion percentages displayed... it let's them more fully understand where the WU is in the program if you report the full value for % complete... but we haven't needed them for a very very long time. (i.e. no more 1% bug!)

Ah, I didn't know that. But if they don't need them...
ID: 36292 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
FluffyChicken
Avatar

Send message
Joined: 1 Nov 05
Posts: 1260
Credit: 369,635
RAC: 0
Message 36373 - Posted: 9 Feb 2007, 7:18:11 UTC - in response to Message 36292.  

The project team actually has embedded some special meanings in to the completion percentages displayed... it let's them more fully understand where the WU is in the program if you report the full value for % complete... but we haven't needed them for a very very long time. (i.e. no more 1% bug!)

Ah, I didn't know that. But if they don't need them...

There is actually no reason why they cannot have both and they could report the time based to BOINC and then use the other either as an addition in the graphics or with a debug switch in the (rosetta) command line when they needed, say at ralph.


Team mauisun.org
ID: 36373 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Christoph

Send message
Joined: 10 Dec 05
Posts: 57
Credit: 1,512,386
RAC: 0
Message 36885 - Posted: 16 Feb 2007, 16:23:51 UTC

No comments of the developers?
ID: 36885 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Chu

Send message
Joined: 23 Feb 06
Posts: 120
Credit: 112,439
RAC: 0
Message 36924 - Posted: 17 Feb 2007, 18:25:15 UTC - in response to Message 36885.  

We will discuss about your suggestion and come back with a proposal on how to make this feather best reflect what is actually happening in Rosetta.
No comments of the developers?


ID: 36924 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : [Request] Progress Estimation



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