Advanced search

Message boards : Number crunching : No tasks available for ACEMD short runs cuda 3.2

Author Message
Profile microchip
Avatar
Send message
Joined: 4 Sep 11
Posts: 110
Credit: 326,102,587
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28489 - Posted: 13 Feb 2013 | 14:23:33 UTC

That's the message I get when trying to fetch new tasks... Am I doing something wrong? I've already resetted the project but still get the same. My driver version is 310.19 on Linux
____________

Team Belgium

Profile dskagcommunity
Avatar
Send message
Joined: 28 Apr 11
Posts: 456
Credit: 817,865,789
RAC: 0
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28490 - Posted: 13 Feb 2013 | 14:40:28 UTC

Hm there are two short queues at the moment i see O.o
____________
DSKAG Austria Research Team: http://www.research.dskag.at



Profile Gattorantolo [Ticino]
Avatar
Send message
Joined: 29 Dec 11
Posts: 44
Credit: 251,211,525
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwat
Message 28492 - Posted: 13 Feb 2013 | 14:44:17 UTC - in response to Message 28489.

I dont know...there are more than 5000 short WU ready to send...
____________
Member of Boinc Italy.

Profile microchip
Avatar
Send message
Joined: 4 Sep 11
Posts: 110
Credit: 326,102,587
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28493 - Posted: 13 Feb 2013 | 14:45:03 UTC

Yes, and I just selected the 4.2 queue and got a task. However, server status reports plenty of 3.2 CUDA tasks to send out, yet I don't get any
____________

Team Belgium

Profile microchip
Avatar
Send message
Joined: 4 Sep 11
Posts: 110
Credit: 326,102,587
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28507 - Posted: 14 Feb 2013 | 9:11:12 UTC
Last modified: 14 Feb 2013 | 9:12:19 UTC

I stopped getting 4.2 CUDA tasks. Now it says the same message that there are no WUs available for short runs but server status shows plenty of them to send out.
____________

Team Belgium

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 28512 - Posted: 14 Feb 2013 | 11:35:00 UTC - in response to Message 28507.

This is due to the changes/deprecation on the 3.1app.

For now chose both short runs,

ACEMD short runs, CUDA 4.2 Yes
ACEMD short runs, CUDA 3.2 Yes
ACEMD beta
ACEMD long runs (8-12 hours on fastest GPU), CUDA 4.2

Thanks,
____________
FAQ's

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

Profile microchip
Avatar
Send message
Joined: 4 Sep 11
Posts: 110
Credit: 326,102,587
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28513 - Posted: 14 Feb 2013 | 11:39:00 UTC

Thanks skgiven

I have only selected the 4.2 queue as the 3.2 one keeps crashing on me... Got a 4.2 task already and am crunching it now :)
____________

Team Belgium

Profile microchip
Avatar
Send message
Joined: 4 Sep 11
Posts: 110
Credit: 326,102,587
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28527 - Posted: 15 Feb 2013 | 13:27:14 UTC

Guys,

I'm still getting "no tasks available" for both the 4.2 and 3.2 queue. What am I doing wrong?
____________

Team Belgium

Profile microchip
Avatar
Send message
Joined: 4 Sep 11
Posts: 110
Credit: 326,102,587
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28529 - Posted: 15 Feb 2013 | 15:10:54 UTC

15-Feb-2013 16:06:12 [GPUGRID] No tasks sent
15-Feb-2013 16:06:12 [GPUGRID] No tasks are available for Short runs (2-3 hours on fastest card)
15-Feb-2013 16:06:12 [GPUGRID] No tasks are available for Short runs (2-3 hours on fastest card)
15-Feb-2013 16:06:12 [GPUGRID] No tasks are available for the applications you have selected.

____________

Team Belgium

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 28530 - Posted: 15 Feb 2013 | 15:28:56 UTC - in response to Message 28529.
Last modified: 15 Feb 2013 | 15:33:49 UTC

It's been like that since yesterday. The Server Status page doesn't correctly reflect the tasks in the respective application selection area. The 3.1apps are being deprecated, so the app selection will change. My guess is that choosing both short runs will result in you being able to run the short tasks after the 3.1app is deprecated and the queue cleared...

For now run the long tasks, or Beta test. The long tasks are the only reliable queue.
____________
FAQ's

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

Profile microchip
Avatar
Send message
Joined: 4 Sep 11
Posts: 110
Credit: 326,102,587
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 28532 - Posted: 15 Feb 2013 | 15:37:08 UTC

Thanks skgiven :)

I tried running the long queue but it errors out in a few minutes. I'm not sure if this is specific to NOELIA tasks or not... For now, I'd like to run only the short ones and I'm willing to wait until the short queue settles down.
____________

Team Belgium

Dylan
Send message
Joined: 16 Jul 12
Posts: 98
Credit: 386,043,752
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwat
Message 29527 - Posted: 21 Apr 2013 | 21:20:25 UTC

I'm a bit confused. Will there ever be Cuda 3.2 short tasks again? Or are they removed? I'm trying to set up a GT 9600m from an old laptop to run GPUGRID, however I get this message from the Event Viewer. Is the GPU to old to run on GPUGRID?


4/21/2013 5:14:15 PM | GPUGRID | update requested by user
4/21/2013 5:14:17 PM | GPUGRID | Sending scheduler request: Requested by user.
4/21/2013 5:14:17 PM | GPUGRID | Requesting new tasks for NVIDIA
4/21/2013 5:14:19 PM | GPUGRID | Scheduler request completed: got 0 new tasks
4/21/2013 5:14:19 PM | GPUGRID | No tasks sent
4/21/2013 5:14:19 PM | GPUGRID | No tasks are available for Short runs (2-3 hours on fastest card)
4/21/2013 5:14:19 PM | GPUGRID | No tasks are available for Short runs (2-3 hours on fastest card)

Jim1348
Send message
Joined: 28 Jul 12
Posts: 819
Credit: 1,591,285,971
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 29529 - Posted: 22 Apr 2013 | 7:02:01 UTC - in response to Message 28532.

I tried running the long queue but it errors out in a few minutes. I'm not sure if this is specific to NOELIA tasks or not...

I now get only errors on the latest NEOLIA work units also on my GTX 650 Ti, on an E8400 Core2 Duo PC. My guess is that there is a minimum hardware configuration required. I will be building a new Ivy Bridge PC shortly, and will put a couple of GTX 660s on it to see if that fixes it.

Simba123
Send message
Joined: 5 Dec 11
Posts: 147
Credit: 69,970,684
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwat
Message 29531 - Posted: 22 Apr 2013 | 8:36:12 UTC - in response to Message 29529.

I tried running the long queue but it errors out in a few minutes. I'm not sure if this is specific to NOELIA tasks or not...

I now get only errors on the latest NEOLIA work units also on my GTX 650 Ti, on an E8400 Core2 Duo PC. My guess is that there is a minimum hardware configuration required. I will be building a new Ivy Bridge PC shortly, and will put a couple of GTX 660s on it to see if that fixes it.


If you can afford it, a 660Ti is a much much better card than the plain 660.

Essentially a 660Ti is a 670 with a slightly smaller memory Bus (148 vs 192 Gb/S)

I'd go as far as to say you would be better off with 1 660Ti Vs 2 660.

Jim1348
Send message
Joined: 28 Jul 12
Posts: 819
Credit: 1,591,285,971
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 29534 - Posted: 22 Apr 2013 | 12:00:49 UTC - in response to Message 29531.
Last modified: 22 Apr 2013 | 12:01:24 UTC

If you can afford it, a 660Ti is a much much better card than the plain 660.

Essentially a 660Ti is a 670 with a slightly smaller memory Bus (148 vs 192 Gb/S)

I'd go as far as to say you would be better off with 1 660Ti Vs 2 660.


With the support of a full Ivy Bridge core (i7-3770), it takes 5 hours 45 minutes to complete a Nathan long on each of my GTX 660s. From what I hear, it is about 5 hours on a GTX 660 Ti. So two 660s appear to be much better, though I don't know how that scales for the Noelias.

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 29537 - Posted: 22 Apr 2013 | 19:06:01 UTC - in response to Message 29534.

Two GTX660's would do more work overall, but a GTX660Ti is ~50% faster than one GTX660. I would still consider going with one GTX660Ti, as that leaves the door open for another a bit further down the line.

Jim1348's NATHAN_dhfr WU's:
I19R14-NATHAN_dhfr36_5-9-32-RND2383_0 4367588 15 Apr 2013 | 17:22:22 UTC 16 Apr 2013 | 2:58:31 UTC Completed and validated 33,155.43 31,863.05 70,800.00 Long runs (8-12 hours on fastest card) v6.18 (cuda42)
I96R5-NATHAN_dhfr36_5-7-32-RND7548_0 4366190 15 Apr 2013 | 8:05:26 UTC 15 Apr 2013 | 17:37:50 UTC Completed and validated 33,198.45 32,000.92 70,800.00 Long runs (8-12 hours on fastest card) v6.18 (cuda42)
I61R6-NATHAN_dhfr36_5-5-32-RND2951_0 4365064 14 Apr 2013 | 22:53:30 UTC 15 Apr 2013 | 8:19:57 UTC Completed and validated 33,165.82 31,779.29 70,800.00 Long runs (8-12 hours on fastest card) v6.18 (cuda42)
I89R14-NATHAN_dhfr36_5-3-32-RND0818_0 4363849 14 Apr 2013 | 13:27:15 UTC 14 Apr 2013 | 23:06:52 UTC Completed and validated 33,179.17 31,820.82 70,800.00 Long runs (8-12 hours on fastest card) v6.18 (cuda42)
I7R1-NATHAN_dhfr36_5-5-32-RND4999_0 4362558 14 Apr 2013 | 4:05:17 UTC 14 Apr 2013 | 13:39:31 UTC Completed and validated 32,790.22 31,956.27 70,800.00 Long runs (8-12 hours on fastest card) v6.18 (cuda42)

On my GTX660Ti:
valx71-NOELIA_PEPTGPRC-1-2-RND0286_1 4382923 21 Apr 2013 | 15:46:37 UTC 22 Apr 2013 | 11:28:04 UTC Completed and validated 33,072.59 31,534.30 113,250.00 Long runs (8-12 hours on fastest card) v6.18 (cuda42)
I2R17-NATHAN_dhfr36_5-14-32-RND1732_2 4382599 21 Apr 2013 | 9:03:57 UTC 21 Apr 2013 | 15:36:20 UTC Completed and validated 21,809.45 18,814.06 70,800.00 Long runs (8-12 hours on fastest card) v6.18 (cuda42)
valx7-NOELIA_PEPTGPRC-0-2-RND1488_2 4367572 21 Apr 2013 | 0:10:23 UTC 21 Apr 2013 | 9:17:44 UTC Completed and validated 31,469.07 31,395.62 113,250.00 ACEMD beta version v6.49 (cuda42)
____________
FAQ's

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

Jim1348
Send message
Joined: 28 Jul 12
Posts: 819
Credit: 1,591,285,971
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 29539 - Posted: 23 Apr 2013 | 7:47:22 UTC - in response to Message 29537.
Last modified: 23 Apr 2013 | 8:42:57 UTC

skgiven,

Thanks for the input. I hadn't seen such a good comparison. But those number for me were on a GTX 650 Ti supported by a Core2 Due E8400 at 3 GHz, not my GTX 660s which were from a few weeks earlier and with the above-noted i7-3770. The 660s ran Nathans at 6 hours plus a minute or two (around 22,000 sec) on a real core, and about 15 minutes longer on a virtual core (27,400 sec), not overclocked. (A small overclock might work, but they crash eventually so I avoid it. The GTX 650 Ti came factory overclocked, and I had to use MSI Afterburner to get it stable.) The Noelias (from the earlier set, not the present ones) were considerably longer, and I think maybe 9 to 10 hours on the 660, though I don't remember the exact numbers now.

My limitations on this PC are heat and fan noise, not so much cost. Originally I had to worry about card length also, but am switching cases now with the new Ivy Bridge build, and agree that the 660 Ti is a nice card. This project is one of the few that uses Kepler cards efficiently.

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 29579 - Posted: 27 Apr 2013 | 12:18:55 UTC

GTX660Ti provides 30% more raw horse power than a GTX660. This should usually translate nicely into 30% better performance at BOINC (unless CPU limits occur). Average game power consumption, on the other hand, is 115 W to 138 W, so an increase of 20% - the GTX660Ti is more power efficient!

MrS
____________
Scanning for our furry friends since Jan 2002

candido
Send message
Joined: 12 Jun 11
Posts: 12
Credit: 150,069,999
RAC: 0
Level
Ile
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 29604 - Posted: 29 Apr 2013 | 22:41:08 UTC

I was having same problem, not being able to download any short runs (any of the two cuda versions) but just got one cuda 4.2...
One, not the usual two wu though
candido

Simba123
Send message
Joined: 5 Dec 11
Posts: 147
Credit: 69,970,684
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwat
Message 29609 - Posted: 30 Apr 2013 | 2:36:22 UTC - in response to Message 29604.
Last modified: 30 Apr 2013 | 2:37:59 UTC

3.1 queue is now closed.

see herehttp://www.gpugrid.net/forum_thread.php?id=3351

<edit> I see you referring to the 3.2 queue, not sure what's up with that, availability in all queues has been a bit sketchy for a couple of days now.

Post to thread

Message boards : Number crunching : No tasks available for ACEMD short runs cuda 3.2

//