No new work for my CPU


log in

Advanced search

Message boards : Number crunching : No new work for my CPU

Author Message
Profile AStrong9
Send message
Joined: 23 May 99
Posts: 4
Credit: 8,700,365
RAC: 0
United States
Message 1201601 - Posted: 2 Mar 2012, 5:52:02 UTC

I have two computers, both have NVIDIA CUDA cards and AMD quad processors. One computer continues to receive a generous amount of CUDA and non-CUDA work. The other has mysteriously stopped receiving non-CUDA work units while the GPU gets heavily overloaded with CUDA work units. The AMD quad CPU sits idling with no work units to crunch. Anybody got any suggestions as to what gives?
____________

bill
Send message
Joined: 16 Jun 99
Posts: 859
Credit: 22,256,645
RAC: 19,327
United States
Message 1201602 - Posted: 2 Mar 2012, 6:01:24 UTC - in response to Message 1201601.

Have you tried changing the settings on your account web page
to cpu only?

Profile AStrong9
Send message
Joined: 23 May 99
Posts: 4
Credit: 8,700,365
RAC: 0
United States
Message 1201605 - Posted: 2 Mar 2012, 6:29:28 UTC - in response to Message 1201602.

I just did. It suspends GPU work, but no CPU work units are downloaded.
____________

Profile AStrong9
Send message
Joined: 23 May 99
Posts: 4
Credit: 8,700,365
RAC: 0
United States
Message 1201607 - Posted: 2 Mar 2012, 6:33:42 UTC

When I do network communications, the project only attempts to get GPU tasks:

3/2/2012 12:18:13 AM SETI@home Sending scheduler request: To fetch work.
3/2/2012 12:18:13 AM SETI@home Requesting new tasks for GPU
3/2/2012 12:18:15 AM SETI@home Scheduler request completed: got 0 new tasks
3/2/2012 12:18:15 AM SETI@home Message from server: No tasks sent
3/2/2012 12:18:15 AM SETI@home Message from server: This computer has reached a limit on tasks in progress
3/2/2012 12:23:23 AM SETI@home Sending scheduler request: To fetch work.
3/2/2012 12:23:23 AM SETI@home Requesting new tasks for GPU
3/2/2012 12:23:25 AM SETI@home Scheduler request completed: got 0 new tasks
3/2/2012 12:23:25 AM SETI@home Message from server: No tasks sent
3/2/2012 12:23:25 AM SETI@home Message from server: This computer has reached a limit on tasks in progress

____________

msattler
Volunteer tester
Avatar
Send message
Joined: 9 Jul 00
Posts: 38149
Credit: 555,426,232
RAC: 611,047
United States
Message 1201612 - Posted: 2 Mar 2012, 7:14:21 UTC - in response to Message 1201597.

I have two computers, both have NVIDIA CUDA cards and AMD quad processors. One computer continues to receive a generous amount of CUDA and non-CUDA work. The other has mysteriously stopped receiving non-CUDA work units while the GPU gets heavily overloaded with CUDA work units. The AMD quad CPU sits idling with no work units to crunch. Anybody got any suggestions as to what gives?

I have had that happen before. When work is scarce, the scheduler apparently tries to always feed the fastest resource first. Usually the GPU. If the GPU does not get up to quota, the CPU gets nothing. And runs out of work.

Work is flowing pretty good right now, I would be surprised if your GPU cache does not hit the limits and then some CPU work starts coming.
____________
*********************************************
Embrace your inner kitty...ya know ya wanna!

I have met a few friends in my life.
Most were cats.

Kevin Olley
Send message
Joined: 3 Aug 99
Posts: 368
Credit: 35,171,401
RAC: 4,228
United Kingdom
Message 1201623 - Posted: 2 Mar 2012, 7:38:29 UTC

Check your preferences, have you got anything in your "connect about every" if so reduce it to 0 days.


____________
Kevin


bill
Send message
Joined: 16 Jun 99
Posts: 859
Credit: 22,256,645
RAC: 19,327
United States
Message 1201650 - Posted: 2 Mar 2012, 9:01:15 UTC - in response to Message 1201605.

I just did. It suspends GPU work, but no CPU work units are downloaded.


Now go into your Boinc Manager , Activity tab and check Use
GPU never. Wait 5 minutes then request work and see what happens.

Profile Gundolf Jahn
Send message
Joined: 19 Sep 00
Posts: 3184
Credit: 356,746
RAC: 15
Germany
Message 1201651 - Posted: 2 Mar 2012, 9:02:49 UTC - in response to Message 1201605.

I just did. It suspends GPU work, but no CPU work units are downloaded.

You did the wrong thing. ;-)

Setting your SETI@home preferences to "Use NVIDIA GPU no" would never suspend GPU work already downloaded. It'll only prevent new CUDA tasks to be assigned to your machine(s).

Gruß,
Gundolf

bill
Send message
Joined: 16 Jun 99
Posts: 859
Credit: 22,256,645
RAC: 19,327
United States
Message 1201654 - Posted: 2 Mar 2012, 9:09:35 UTC - in response to Message 1201651.

I just did. It suspends GPU work, but no CPU work units are downloaded.

You did the wrong thing. ;-)

Setting your SETI@home preferences to "Use NVIDIA GPU no" would never suspend GPU work already downloaded. It'll only prevent new CUDA tasks to be assigned to your machine(s).

Gruß,
Gundolf


Not the wrong thing if your trying to force cpu work.
First you tell the system that you only want cpu work, then you
set your computer to only work on cpu work units.

The message he's getting that says "3/2/2012 12:18:15 AM SETI@home Message from server: This computer has reached a limit on tasks in progress" tells the tale.


bill
Send message
Joined: 16 Jun 99
Posts: 859
Credit: 22,256,645
RAC: 19,327
United States
Message 1201655 - Posted: 2 Mar 2012, 9:11:59 UTC - in response to Message 1201607.

It would help if you unhid your computers.

LadyL
Volunteer tester
Avatar
Send message
Joined: 14 Sep 11
Posts: 1679
Credit: 5,230,097
RAC: 0
Message 1201660 - Posted: 2 Mar 2012, 9:55:53 UTC - in response to Message 1201597.

I have two computers, both have NVIDIA CUDA cards and AMD quad processors. One computer continues to receive a generous amount of CUDA and non-CUDA work. The other has mysteriously stopped receiving non-CUDA work units while the GPU gets heavily overloaded with CUDA work units. The AMD quad CPU sits idling with no work units to crunch. Anybody got any suggestions as to what gives?


it helps to track problems if you unhide your computers.

When you're not getting work, the first question is 'is it not asking or is it not receiving?' - check your massages tab/event log and post.

It can have all sorts of reasons, but as Matk stated with the current server status you should be getting work now, so if you don't, I suspect you somehow disabled CPU work.

Claggy
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 4039
Credit: 32,691,076
RAC: 800
United Kingdom
Message 1201676 - Posted: 2 Mar 2012, 11:57:40 UTC - in response to Message 1201660.

check your massages tab/event log and post.

We have a massages tab now do we? ;-)

Claggy

LadyL
Volunteer tester
Avatar
Send message
Joined: 14 Sep 11
Posts: 1679
Credit: 5,230,097
RAC: 0
Message 1201677 - Posted: 2 Mar 2012, 12:11:28 UTC

Sorry I was answering the wrong copy of you post, didn't see you'd opened a fresh thread.

From your log you can see it's not requesting.

In the preferences page Gundolf quoted, check that you have CPU work enabled.
If you run other projects on that host, it may be a resource thing, but then the CPU should be doing other projects not sitting idle.

We may have to delve into workfetch debug logs...

Richard Haselgrove
Volunteer tester
Send message
Joined: 4 Jul 99
Posts: 8373
Credit: 46,532,119
RAC: 13,295
United Kingdom
Message 1201678 - Posted: 2 Mar 2012, 12:13:44 UTC - in response to Message 1201676.

check your massages tab/event log and post.

We have a massages tab now do we? ;-)

Claggy

That's another reason for asking him to unhide his computers. As things stand, we don't know whether he's using BOINC v6.10, v6.12 - or anything else. It's hard to tailor specific advice for a new visitor to these boards without that information.

LadyL
Volunteer tester
Avatar
Send message
Joined: 14 Sep 11
Posts: 1679
Credit: 5,230,097
RAC: 0
Message 1201679 - Posted: 2 Mar 2012, 12:16:32 UTC - in response to Message 1201676.

check your massages tab/event log and post.

We have a massages tab now do we? ;-)

Claggy


I don't know about you, but I could do with one, I've got a stiff neck.

Profile ivan
Volunteer tester
Avatar
Send message
Joined: 5 Mar 01
Posts: 578
Credit: 130,634,243
RAC: 112,812
United Kingdom
Message 1201838 - Posted: 2 Mar 2012, 20:09:21 UTC - in response to Message 1201679.

check your massages tab/event log and post.

We have a massages tab now do we? ;-)

I don't know about you, but I could do with one, I've got a stiff neck.

Me too, but mine's from a rather spectacular bike-racing incident (not the one that led to the tyre rubber on my number plate in my avatar photo, that one didn't even bring me down...).
____________

Message boards : Number crunching : No new work for my CPU

Copyright © 2014 University of California