Advanced search

Message boards : Number crunching : 2x72-SDOERR_OPMamber6-0-1-RND1929 - Deadline Warning ?!

Author Message
Profile Sutaru Tsureku
Send message
Joined: 1 May 09
Posts: 13
Credit: 3,655,193
RAC: 0
Level
Ala
Scientific publications
watwatwat
Message 45625 - Posted: 8 Dec 2016 | 19:15:38 UTC
Last modified: 8 Dec 2016 | 19:26:31 UTC

2x72-SDOERR_OPMamber6-0-1-RND1929

My PC with NV GT730 got a copy of this above WU.

(The estimated times are ~ correct in BOINC.)
BOINC said this above WU will last 25 days...

After 10 hrs calculation time just 2.5 % progress.

The CPU usage was very low, the VGA VRAM usage was very high ~ 500 MB.
IIRC normally it's just ~ 1xx or 2xx MB.

I needed to abort this WU... - because upload and report 20 days after deadline would be nonsense...

Is there something wrong with this WU?

Thanks.
____________

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45627 - Posted: 8 Dec 2016 | 19:31:18 UTC - in response to Message 45625.

Is there something wrong with this WU?

Yes, it's huge. Abort it. With a GT730 I'd recommend running the short WUs or the shorter variety long WUs. Aborting a WU harms nothing. In fact in this case it helps, as it will then be assigned to another host (hopefully one fast enough to complete it). We've asked many times that a "super Long WU" queue be added, but so far admin ears are deaf to our pleas.

PappaLitto
Send message
Joined: 21 Mar 16
Posts: 508
Credit: 4,607,207,341
RAC: 0
Level
Arg
Scientific publications
watwatwatwatwatwatwatwat
Message 45639 - Posted: 9 Dec 2016 | 11:37:32 UTC - in response to Message 45625.

2x72-SDOERR_OPMamber6-0-1-RND1929

My PC with NV GT730 got a copy of this above WU.

Thanks.


People with slow gpus can still help, just go into project settings on the GPUGrid website under account and deselect long WU and only have short ones.

Profile koschi
Avatar
Send message
Joined: 14 Aug 08
Posts: 124
Credit: 188,684,198
RAC: 0
Level
Ile
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 45645 - Posted: 9 Dec 2016 | 21:01:34 UTC - in response to Message 45627.
Last modified: 9 Dec 2016 | 21:03:05 UTC

We've asked many times that a "super Long WU" queue be added, but so far admin ears are deaf to our pleas.


I'd also love to get just these huge ones.
Just completed a 24h WU on my GTX 1060 3GB, it yielded 903k credits, wow!

Bedrich Hajek
Send message
Joined: 28 Mar 09
Posts: 404
Credit: 5,832,269,764
RAC: 1,788,981
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45649 - Posted: 10 Dec 2016 | 5:39:00 UTC - in response to Message 45645.

We've asked many times that a "super Long WU" queue be added, but so far admin ears are deaf to our pleas.


I'd also love to get just these huge ones.
Just completed a 24h WU on my GTX 1060 3GB, it yielded 903k credits, wow!



And, so do I !!


Name 4nef-SDOERR_OPMamber6-0-1-RND6739_2
Workunit 12185716
Created 9 Dec 2016 | 9:15:02 UTC
Sent 9 Dec 2016 | 9:15:11 UTC
Received 10 Dec 2016 | 5:02:31 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 263612
Report deadline 14 Dec 2016 | 9:15:11 UTC
Run time 70,428.08
CPU time 70,218.94
Validate state Valid
Credit 1,089,669.00
Application version Long runs (8-12 hours on fastest card) v8.48 (cuda65)

http://www.gpugrid.net/result.php?resultid=15822755



And on my other computer, I ran these two simultaneously:

15818621 12185601 8 Dec 2016 | 8:31:08 UTC 9 Dec 2016 | 17:18:24 UTC Completed and validated 111,374.20 108,406.50 914,707.50 Long runs (8-12 hours on fastest card) v8.48 (cuda65)
15818608 12185595 8 Dec 2016 | 8:31:08 UTC 9 Dec 2016 | 16:37:21 UTC Completed and validated 107,217.12 103,787.30 373,065.00 Long runs (8-12 hours on fastest card) v8.48 (cuda65)

http://www.gpugrid.net/result.php?resultid=15818621

http://www.gpugrid.net/result.php?resultid=15818608

They are a lot more challenging and fun to run, then the short units pretending that they are long units!



Stefan
Volunteer moderator
Project developer
Project scientist
Send message
Joined: 5 Mar 13
Posts: 348
Credit: 0
RAC: 0
Level

Scientific publications
wat
Message 45650 - Posted: 10 Dec 2016 | 10:58:49 UTC

I was a tad bit worried because I ran the tests for these WUs on a 1080 so the projected runtime was a bit optimistic compared to the previous ones :P But it's a one-off deal again so don't worry, they'll be over in a few days

Profile koschi
Avatar
Send message
Joined: 14 Aug 08
Posts: 124
Credit: 188,684,198
RAC: 0
Level
Ile
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 45653 - Posted: 10 Dec 2016 | 15:01:08 UTC

We want more, not less :-D

Bedrich Hajek
Send message
Joined: 28 Mar 09
Posts: 404
Credit: 5,832,269,764
RAC: 1,788,981
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45655 - Posted: 10 Dec 2016 | 16:40:32 UTC - in response to Message 45653.

We want more, not less :-D


I agree!




Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45659 - Posted: 11 Dec 2016 | 16:56:42 UTC - in response to Message 45653.
Last modified: 11 Dec 2016 | 16:59:53 UTC

We want more, not less :-D

I'd agree only if they're put in a 3rd queue so that people with fast GPUs can choose to run them. It's not good at all if slower GPUs are assigned these super long WUs. It's a waste of resources for both the project and user. Maybe if creating a 3rd queue is too difficult, they could at least be assigned to cuda80 GPUs only?

Betting Slip
Send message
Joined: 5 Jan 09
Posts: 670
Credit: 2,498,095,550
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45665 - Posted: 12 Dec 2016 | 8:26:29 UTC - in response to Message 45659.
Last modified: 12 Dec 2016 | 8:28:43 UTC

Maybe if creating a 3rd queue is too difficult, they could at least be assigned to cuda80 GPUs only?


Hold on 980ti, 980 and 970 are nearly just as capable at running these WU's as a 1080 and better than some Pascal cards so stop being elitist. :-)

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45677 - Posted: 12 Dec 2016 | 16:31:07 UTC - in response to Message 45665.
Last modified: 12 Dec 2016 | 16:34:56 UTC

Maybe if creating a 3rd queue is too difficult, they could at least be assigned to cuda80 GPUs only?

Hold on 980ti, 980 and 970 are nearly just as capable at running these WU's as a 1080 and better than some Pascal cards so stop being elitist. :-)

Yes, I realize that. Sure I'm being elitist. I run mostly old 750Ti GPUs as you know. Calm down. Trying to give them an alternative that they actually may be able to do. Right now the super long WUs are being given to all kinds of super slow GPUs that have no hope of completing them in 5 days. It may be easier for them to assign them to cuda80 only. So far all the 10xx GPUs are at least fast enough to run them. A MUCH BETTER solution would be a 3rd queue but I don't think they know how to set one up. As you just mentioned in the other thread, there must be some IT students that would be willing to give them a hand with server configuration.

A 3rd queue would allow the 980Ti, 980, 1080, 1070 and 970 crowd to efficiently run the super long WUs if they choose to add the super long queue in project preferences.

PappaLitto
Send message
Joined: 21 Mar 16
Posts: 508
Credit: 4,607,207,341
RAC: 0
Level
Arg
Scientific publications
watwatwatwatwatwatwatwat
Message 45680 - Posted: 12 Dec 2016 | 18:28:25 UTC - in response to Message 45677.

A 3rd queue would allow the 980Ti, 980, 1080, 1070 and 970 crowd to efficiently run the super long WUs if they choose to add the super long queue in project preferences.


+1

Profile koschi
Avatar
Send message
Joined: 14 Aug 08
Posts: 124
Credit: 188,684,198
RAC: 0
Level
Ile
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 45694 - Posted: 13 Dec 2016 | 12:43:50 UTC - in response to Message 45677.

[quote][quote]A 3rd queue would allow the 980Ti, 980, 1080, 1070 and 970 crowd to efficiently run the super long WUs if they choose to add the super long queue in project preferences.



Hey, now don't exclude the 1060s please :-D Its in between GTX 970 and 980 performance wise...

Erich56
Send message
Joined: 1 Jan 15
Posts: 828
Credit: 3,469,102,479
RAC: 1,335,485
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwat
Message 45697 - Posted: 13 Dec 2016 | 16:11:35 UTC - in response to Message 45680.

A 3rd queue would allow the 980Ti, 980, 1080, 1070 and 970 crowd to efficiently run the super long WUs if they choose to add the super long queue in project preferences.

+1

+1

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45710 - Posted: 13 Dec 2016 | 20:02:47 UTC - in response to Message 45694.

[quote][quote]A 3rd queue would allow the 980Ti, 980, 1080, 1070 and 970 crowd to efficiently run the super long WUs if they choose to add the super long queue in project preferences.

Hey, now don't exclude the 1060s please :-D Its in between GTX 970 and 980 performance wise...

The above list is just examples. Nothing is excluded. It would simply allow you to choose to run WUs from the "super long" queue if you so desire.

Post to thread

Message boards : Number crunching : 2x72-SDOERR_OPMamber6-0-1-RND1929 - Deadline Warning ?!