Advanced search

Message boards : Number crunching : Unusually LONG run time on NOELIA_hfXA_long file

Author Message
Tyendol
Send message
Joined: 9 Nov 11
Posts: 4
Credit: 498,271,824
RAC: 789,252
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 27954 - Posted: 6 Jan 2013 | 19:53:49 UTC

I know you folks sent out a message saying that you had some long run times that were going to credit for 90K points, and I have done 3 of them already. However, my system is currently crunching on:

2x47_18-NOELIA_hfXA_long-0-2-RND2768_1

The previous files had taken my GTX460 about 20-24h to complete; however, this file being crunched has already taken 35h, and is still only 54% towards completion.

I thought I would mention it, in case anyone else was having the same problem. It's running on high priority now, since it's completion is tomorrow, and still showing it has 26h to go. Whatever you do, don't cancel the task - just let it keep going; but, thought it was odd that it was taking so long...and 90K seemed fine for the other ones; but this bad boy better give me 2-3x that. =D

Thanks for listening.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 27955 - Posted: 6 Jan 2013 | 20:48:32 UTC - in response to Message 27954.
Last modified: 6 Jan 2013 | 20:55:57 UTC

Your task is running using the Cuda3.1 application,


    Long runs (8-12 hours on fastest card) v6.16 (cuda31)


The 3.1 app is slower for Fermi and Kepler cards and was deprecated (removed) from the Long queue about 11h after you downloaded that task. So you managed to pick up one of the last 3.1 tasks, as did I. The results are still useful, so you can let it finish.
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

flashawk
Send message
Joined: 18 Jun 12
Posts: 297
Credit: 3,572,627,986
RAC: 0
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwat
Message 27956 - Posted: 6 Jan 2013 | 21:01:43 UTC
Last modified: 6 Jan 2013 | 21:02:39 UTC

I just aborted one that I downloaded a couple hours ago, I stopped looking for them thinking they were all gone and cought it when looking at the GPU load.

Tyendol
Send message
Joined: 9 Nov 11
Posts: 4
Credit: 498,271,824
RAC: 789,252
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 27957 - Posted: 6 Jan 2013 | 21:03:58 UTC - in response to Message 27955.

Thanks for the info. Will definitely allow it to complete; it's football day anyway. :)

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 27958 - Posted: 7 Jan 2013 | 1:04:54 UTC - in response to Message 27956.

The 3.1 app is still used in the short (normal length) queue.
It's only been deprecated from the Long queue.

____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

flashawk
Send message
Joined: 18 Jun 12
Posts: 297
Credit: 3,572,627,986
RAC: 0
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwat
Message 27960 - Posted: 7 Jan 2013 | 4:06:18 UTC

Okay, I understand now. I'm going to uncheck those for a while, I'm juggling cats over here trying to keep my upload queue's empty.

mhhall
Send message
Joined: 21 Mar 10
Posts: 23
Credit: 861,667,631
RAC: 0
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 27964 - Posted: 7 Jan 2013 | 16:20:54 UTC

It seems that a side effect of these longer estimated times
may be that the BOINC client can not accept additional GPUgrid
jobs as before. My system has two GPU, and had been accepting
two more jobs (at max).

With this extended time estimate (on my systems in the 300+ hour
range when initially received), it appears that the client thinks
that it won't complete any additional jobs in sufficient time.

This may be one of those things where we've moved too far in
the new direction.

Post to thread

Message boards : Number crunching : Unusually LONG run time on NOELIA_hfXA_long file

//