No work?


log in

Advanced search

Message boards : Number crunching : No work?

1 · 2 · Next
Author Message
Wayne Miller
Avatar
Send message
Joined: 15 Mar 03
Posts: 13
Credit: 199,410
RAC: 4
United States
Message 1285937 - Posted: 20 Sep 2012, 18:02:37 UTC

I noticed yesterday that all of my work units for Seti have finished and no new work units have downloaded. Is there really no new work for me, or has something gone awry?
____________

Profile Area 51
Avatar
Send message
Joined: 31 Jan 04
Posts: 965
Credit: 42,193,520
RAC: 0
United Kingdom
Message 1285977 - Posted: 20 Sep 2012, 20:14:10 UTC - in response to Message 1285937.

I noticed yesterday that all of my work units for Seti have finished and no new work units have downloaded. Is there really no new work for me, or has something gone awry?



There is work, but the network is maxed out. Patience is the name of the game - 'all good things come to those who wait'..... Your work requests may simply be lost in the sea of hosts shouting for work.

Seriously, you just have to let your client keep asking - sooner or later it'll get some work. You may like to review your cache settings to ensure you keep some tasks in hand. The task deadlines are usually quite generous, so holding some tasks in the cupboard for when the network gremlins strike is not a bad idea...
____________

Wayne Miller
Avatar
Send message
Joined: 15 Mar 03
Posts: 13
Credit: 199,410
RAC: 4
United States
Message 1285979 - Posted: 20 Sep 2012, 20:18:42 UTC

Thank you for your response. I will up my cache settings.

____________

Profile Fred E.Project donor
Volunteer tester
Send message
Joined: 22 Jul 99
Posts: 768
Credit: 24,139,004
RAC: 0
United States
Message 1285987 - Posted: 20 Sep 2012, 20:27:23 UTC

You may have to watch to see that the cache is not filling up with WCG tasks. I have that problem when running WCG with BOINC 7, and have to use No New Tasks to keep the two projects balanced. Also, what cache settings are you using?

There's very few Astropulse tasks available today for your ATI 5700, but plenty of Multibeam for your CPU.
____________
Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.

Wayne Miller
Avatar
Send message
Joined: 15 Mar 03
Posts: 13
Credit: 199,410
RAC: 4
United States
Message 1286267 - Posted: 21 Sep 2012, 14:41:33 UTC
Last modified: 21 Sep 2012, 14:43:07 UTC

I had my cache settings at 2 days and 1 additional day, but now I just raised them to 3 and 3. Everything seemed to be working all right until 9-19. Thats when the final seti wu uploaded.
____________

Scrmndmn
Send message
Joined: 26 Jul 99
Posts: 1
Credit: 2,402,432
RAC: 1,700
United States
Message 1288212 - Posted: 26 Sep 2012, 20:55:28 UTC - in response to Message 1286267.

I'm having the same issue. Around the same time last week as Wayne my pc stopped processing files on the cpu and was just running on my vid card. Now I'm just getting nothing to process in either spot. I'll be patient and see what happens.

Spookwoman
Send message
Joined: 3 Oct 11
Posts: 9
Credit: 1,456,610
RAC: 60
Ireland
Message 1288255 - Posted: 27 Sep 2012, 0:27:31 UTC

Glad its not just me. 2 computer upgrades today. Decommissioned old pc and got the next one in line working. Its just updated not problems but the brand new pc is not downloading.

Spookwoman
Send message
Joined: 3 Oct 11
Posts: 9
Credit: 1,456,610
RAC: 60
Ireland
Message 1288256 - Posted: 27 Sep 2012, 0:28:53 UTC - in response to Message 1288255.

Now its decided to download a task.

parl
Send message
Joined: 22 May 04
Posts: 85
Credit: 1,517,563
RAC: 488
United States
Message 1288687 - Posted: 28 Sep 2012, 5:18:43 UTC

I abandoned my old computer which has been working on SETI@home for many years. I now have set up 2 new (to me) computers and both of them are processing work units, much faster than the venerable box they replaced.

I, too, am having problems getting work. Often, if I let the program get work unattended (which seems to happen at 3:30 AM local/Pacific time), there's none available. In a larger sense, I guess this is due to the flurry of short work units which were distributed for a while, but seem to be past.

I think I'm only using local preferences. Am I able to set a cache size locally, or is that feature only available for preferences set at the SETI@home site?

Thanks for your support.

Jam0r
Send message
Joined: 3 Apr 99
Posts: 3
Credit: 2,702,560
RAC: 0
United Kingdom
Message 1288731 - Posted: 28 Sep 2012, 8:43:40 UTC

I have set 10 + 10 days and still run out of WUs :(
____________

Richard HaselgroveProject donor
Volunteer tester
Send message
Joined: 4 Jul 99
Posts: 8813
Credit: 53,503,100
RAC: 45,596
United Kingdom
Message 1288745 - Posted: 28 Sep 2012, 9:29:07 UTC - in response to Message 1288731.

I have set 10 + 10 days and still run out of WUs :(

If you try for a 20 day cache - which implies a 20 day turnround - then you'll never complete the current batch of 14-day deadline tasks in time, and the server won't send you any.

Ask for less, and you might actually get more.

Profile Fred E.Project donor
Volunteer tester
Send message
Joined: 22 Jul 99
Posts: 768
Credit: 24,139,004
RAC: 0
United States
Message 1288749 - Posted: 28 Sep 2012, 9:51:04 UTC

I think I'm only using local preferences. Am I able to set a cache size locally, or is that feature only available for preferences set at the SETI@home site?

The computing preferences in BOINC include the cache settings and override what you have in computing preferences on the website.

Work is flowing pretty well today, so suggest you experiment with those settings. I find that a very low value for the "max additional work buffer" works best.
____________
Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.

ChrisSibbald
Send message
Joined: 23 Jul 11
Posts: 18
Credit: 23,582,502
RAC: 0
Canada
Message 1288779 - Posted: 28 Sep 2012, 11:41:40 UTC - in response to Message 1288749.

I have tried various download settings on local settings to no avail...very little work for the last 4 days...something is wrong.

Profile Mr. KevvyProject donor
Volunteer tester
Avatar
Send message
Joined: 15 May 99
Posts: 735
Credit: 78,568,289
RAC: 9,131
Canada
Message 1288813 - Posted: 28 Sep 2012, 14:23:19 UTC

There seems to be a shortage of CPU WUs. I have one box at least whose CPU cores are sitting idle (and not for a lack of requesting as I've watched the log) and although it gets GPU WUs whenever it asks, not a single CPU one in nigh two days.
____________
“Never doubt that a small group of thoughtful, committed citizens can change the world; indeed, it's the only thing that ever has.”
--- Margaret Mead

Profile Fred E.Project donor
Volunteer tester
Send message
Joined: 22 Jul 99
Posts: 768
Credit: 24,139,004
RAC: 0
United States
Message 1288814 - Posted: 28 Sep 2012, 14:26:22 UTC

I have tried various download settings on local settings to no avail...very little work for the last 4 days...something is wrong.

I'm getting work on about 1 of 2 requests, althouh sometimes it is only one task. Is BOINC asking for work? If not, some of the common reasons include a suspended task, a file transfer in retry status, or too much work from other projects. Also check your other local prefs to make sure there's not an inadvertent entry restricting network usage.

Those errors on xxx410 look like the sleep bug from that 295.73 driver. See Claggy's sticky thread.
____________
Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.

Profile Fred E.Project donor
Volunteer tester
Send message
Joined: 22 Jul 99
Posts: 768
Credit: 24,139,004
RAC: 0
United States
Message 1288817 - Posted: 28 Sep 2012, 14:32:55 UTC

There seems to be a shortage of CPU WUs. I have one box at least whose CPU cores are sitting idle (and not for a lack of requesting as I've watched the log) and although it gets GPU WUs whenever it asks, not a single CPU one in nigh two days.


Scheduler will keep filling gpu requests first so long as BOINC is asking for GPU work. You can lower the cache setting until BOINC thinks it has enough gpu work and only asks for cpu work. But the quickest fix is to turn off gpu fetch in your S@H preferences (use Nvidia GPU). That only impacts work fetch, not the processing of work on hand. Change it back when you have enough cpu work.
____________
Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.

Profile Mr. KevvyProject donor
Volunteer tester
Avatar
Send message
Joined: 15 May 99
Posts: 735
Credit: 78,568,289
RAC: 9,131
Canada
Message 1288822 - Posted: 28 Sep 2012, 14:44:24 UTC - in response to Message 1288817.


Scheduler will keep filling gpu requests first so long as BOINC is asking for GPU work. You can lower the cache setting until BOINC thinks it has enough gpu work and only asks for cpu work. But the quickest fix is to turn off gpu fetch in your S@H preferences (use Nvidia GPU). That only impacts work fetch, not the processing of work on hand. Change it back when you have enough cpu work.


Very nice tip and thank you. :^)
____________
“Never doubt that a small group of thoughtful, committed citizens can change the world; indeed, it's the only thing that ever has.”
--- Margaret Mead

parl
Send message
Joined: 22 May 04
Posts: 85
Credit: 1,517,563
RAC: 488
United States
Message 1288940 - Posted: 28 Sep 2012, 17:48:44 UTC

I guess my question is: Where do I find the settings for the max additional work buffer mentioned elsewhere in this thread? I'm running BOINC Manager 7.0.28/2.8.10 and my Application is SETI@home Enhanced 6.03, from what I see on the Tasks page.

When I go to Tools | Computing preferences, I've checked the 4 tabs and don't find anything related to getting additional work.

At 3:30 AM (Where is this set?) BOINC requested tasks and was allocated 31 of them. It's spend the rest of the time trying to d/l them, but is still trying for 10 of them. I see that a passel are still the shorties.
____________

Profile Fred E.Project donor
Volunteer tester
Send message
Joined: 22 Jul 99
Posts: 768
Credit: 24,139,004
RAC: 0
United States
Message 1288948 - Posted: 28 Sep 2012, 18:01:29 UTC

I guess my question is: Where do I find the settings for the max additional work buffer mentioned elsewhere in this thread? I'm running BOINC Manager 7.0.28/2.8.10 and my Application is SETI@home Enhanced 6.03, from what I see on the Tasks page.

When I go to Tools | Computing preferences, I've checked the 4 tabs and don't find anything related to getting additional work.


It is the Network usage tab:
Minimum work buffer is the # of days you want for all projects. Try 5 to 10.
Max additional work buffer can cause problems when a large value is set. I use .01 days.

At 3:30 AM (Where is this set?) BOINC requested tasks...

It is getting the local time from the operating system. If it's not correct, have Windows update the time.
____________
Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.

Grant (SSSF)
Send message
Joined: 19 Aug 99
Posts: 5946
Credit: 62,392,236
RAC: 38,483
Australia
Message 1289013 - Posted: 28 Sep 2012, 20:36:05 UTC - in response to Message 1288813.
Last modified: 28 Sep 2012, 21:19:27 UTC

There seems to be a shortage of CPU WUs.

I've got similar issues as a result of the problems getting work from the Scheduler over the last week or so. Because of the lack of work, it fills the resource that can process the most work- my GPU. However the GPU cache is almost full, the CPU i've run out of work several times.

It would be nice if the Scheduler could allocate work based on need, not processing ability.

eg- 5 day cache, CPU has one days worth of work left, GPU has 4. Allocate work to the CPU till it matches the GPUs present cache, then start allocating work to both.
Or maybe better to do this with the client? No need to make the Scheduler more complicated than it already is.


EDIT- just posted a few thoughts on that in the Wish list thread in Questions & Answers.
____________
Grant
Darwin NT.

1 · 2 · Next

Message boards : Number crunching : No work?

Copyright © 2014 University of California