Advanced search

Message boards : Number crunching : New fragxa3 ultralong?

Author Message
Profile dskagcommunity
Avatar
Send message
Joined: 28 Apr 11
Posts: 456
Credit: 817,865,789
RAC: 0
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24514 - Posted: 21 Apr 2012 | 17:38:52 UTC
Last modified: 21 Apr 2012 | 18:37:48 UTC

Is there something wrong with the new MJHARVEY WUs?

It seems to need wide! over! 30h to complete on 560TI (running @ 98% and full standart clockspeed). And that in the short Queue! O.o Aborted now 4 of them..
____________
DSKAG Austria Research Team: http://www.research.dskag.at



Rick A. Sponholz
Avatar
Send message
Joined: 20 Jan 09
Posts: 52
Credit: 2,518,707,115
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24516 - Posted: 21 Apr 2012 | 18:29:49 UTC
Last modified: 21 Apr 2012 | 18:38:24 UTC

I've got the same problem with these MJHARVEY WU's. Four of them have been running 14 hours, and only 28% complete. This is on GTX 295 @ 1081GFLOPS Peak. I'm monitoring the GPU clocking with GPU Shark, and the GPU's have not downclocked. Please let me know what's up with these WU's. Rick
____________

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1617
Credit: 8,169,894,351
RAC: 16,868,276
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24523 - Posted: 21 Apr 2012 | 23:53:21 UTC

And much the same here - my first is at about 15% after 12 hours, on a shared GTX 470.

Even in the count=0.5 configuration, that card can usually do two or three short-queue tasks per day - I think these MJHARVEYs belong in the long queue, at least.

Blizzie
Send message
Joined: 23 Nov 08
Posts: 12
Credit: 3,505,971
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwatwatwatwat
Message 24527 - Posted: 22 Apr 2012 | 6:52:16 UTC

Yup.. also a ~30 hour WU here for me. 12 hours @ 30% completion on a GTX 570. Wow.

GPUGRID Role account
Send message
Joined: 15 Feb 07
Posts: 134
Credit: 1,349,535,983
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 24528 - Posted: 22 Apr 2012 | 9:09:26 UTC - in response to Message 24514.

Sorry all, I made a mistake in their configuration. They're deleted now and will be resubmitted later.

MJH

[DPC]Charley
Send message
Joined: 4 Oct 11
Posts: 2
Credit: 4,380,100
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwatwatwat
Message 24529 - Posted: 22 Apr 2012 | 11:26:04 UTC

Deleted and mistake as in even the one that I have at 74% completion after 28h is worthless?

Profile algabe
Send message
Joined: 23 May 10
Posts: 9
Credit: 799,670,301
RAC: 196,958
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24531 - Posted: 22 Apr 2012 | 11:55:57 UTC

As well, 20 hours per unit spins to the trash, i do not even a little bit of grace.

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1617
Credit: 8,169,894,351
RAC: 16,868,276
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24532 - Posted: 22 Apr 2012 | 12:40:06 UTC - in response to Message 24528.

Sorry all, I made a mistake in their configuration. They're deleted now and will be resubmitted later.

MJH

Are you sure? I've just aborted WU 3359817 (unstarted in my cache, stuck behind a running one after 24 hours), and a replacement task has been created and put on the queue for sending out.

Profile Steve Dodd
Send message
Joined: 26 Dec 08
Posts: 17
Credit: 2,922,937,729
RAC: 13,144,762
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24534 - Posted: 22 Apr 2012 | 13:30:35 UTC
Last modified: 22 Apr 2012 | 13:31:37 UTC

I'm pulling my cards off GPUGRID until this is sorted out. I can't "afford" to waste 30 hours of work for nothing. And, yes, theses are still downloading.

Rick A. Sponholz
Avatar
Send message
Joined: 20 Jan 09
Posts: 52
Credit: 2,518,707,115
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24538 - Posted: 22 Apr 2012 | 15:57:54 UTC - in response to Message 24532.

As of 11:00 Eastern Daylight Time on 22.April.2012, I'm also still getting these Ultra Long MJHArvey WU's. They seem to be wingman WU's (_3) rather than originals. Also note however, I'm also getting NEW MJHARVEY WU's that are NOT ultralong. Moderator, PLEASE LET US KNOW WHAT"S GOING ON! Rick
____________

Shadowlurker
Send message
Joined: 29 Mar 09
Posts: 4
Credit: 152,630,068
RAC: 0
Level
Ile
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24540 - Posted: 22 Apr 2012 | 16:31:44 UTC

I had 2 harvey wus that ran over 24 hours each and errored out. I do not accept long run wu's and these definitely seem to qualify so I don't know why I got them in the first place, but now I have to babysit my computer and abort them manually. Think it's time to move my GPUs to another project til they get this figured out.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 24541 - Posted: 22 Apr 2012 | 16:36:56 UTC - in response to Message 24538.

Some Wus will escape cancelling (for instance all the ones aborted).
I have done another round of cancellations.

Let me know if you still received those.

gdf

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1617
Credit: 8,169,894,351
RAC: 16,868,276
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24544 - Posted: 22 Apr 2012 | 18:11:02 UTC - in response to Message 24541.

Well, mine got the message OK:

22/04/2012 18:33:01 | GPUGRID | Sending scheduler request: Requested by project.
22/04/2012 18:33:04 | GPUGRID | Result 19x45-MJHARVEY_FRAGXA3-0-30-RND6951_0 is no longer usable
22/04/2012 18:33:04 | GPUGRID | [sched_op] Reason: Unrecoverable error for task 19x45-MJHARVEY_FRAGXA3-0-30-RND6951_0 (aborted by project - no longer usable)
22/04/2012 18:33:36 | GPUGRID | Sending scheduler request: To report completed tasks.
22/04/2012 18:33:36 | GPUGRID | Reporting 1 completed tasks, not requesting new tasks
22/04/2012 18:33:38 | GPUGRID | [sched_op] handle_scheduler_reply(): got ack for task 19x45-MJHARVEY_FRAGXA3-0-30-RND6951_0

and since then I've picked up new work from different - normal-sized - tasks.

The fragxa3 would have reached about 38% in 29 hours by then (I'd checked it not long before), but is still showing unreported on the website: task 5268502

Profile dskagcommunity
Avatar
Send message
Joined: 28 Apr 11
Posts: 456
Credit: 817,865,789
RAC: 0
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 24813 - Posted: 8 May 2012 | 18:03:47 UTC

Oh man thats bad, would be interesting when we can compute short units again on unattended machines :/
____________
DSKAG Austria Research Team: http://www.research.dskag.at



Post to thread

Message boards : Number crunching : New fragxa3 ultralong?

//