Welcome to MilkyWay@home

Getting a (small) buffer of GPU units

Message boards : Number crunching : Getting a (small) buffer of GPU units
Message board moderation

To post messages, you must log in.

AuthorMessage
[DPC]Charley

Send message
Joined: 14 Feb 12
Posts: 4
Credit: 869,897
RAC: 0
Message 60095 - Posted: 3 Oct 2013, 17:33:19 UTC
Last modified: 3 Oct 2013, 17:34:12 UTC

Hi guys,

I'm pretty familiar with BOINC but I'm unable to convince it to get multiple Milkyway@home tasks for my GPU. I'd love to do this as my internet connection is a bit dodgy at times and a single unit only runs for 2 and a bit minutes.

I've set the minimum and additional work buffer both to 1 day (but also tried with 2, 3 and 10 in either). I've got a couple of PrimeGrid tasks running on my CPU, but they've only got under an hour left each. There's no lack of drive space, my connection is all ok. All I get is the following message:

13476	Milkyway@Home	10/3/2013 7:30:24 PM	update requested by user	
13477	Milkyway@Home	10/3/2013 7:30:26 PM	Sending scheduler request: Requested by user.	
13478	Milkyway@Home	10/3/2013 7:30:26 PM	Not reporting or requesting tasks	
13479	Milkyway@Home	10/3/2013 7:30:27 PM	Scheduler request completed	


ANy idea what I can do to get some more tasks? Say 10 or 20 would be enough.
ID: 60095 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,938,381
RAC: 22,848
Message 60103 - Posted: 4 Oct 2013, 12:27:47 UTC - in response to Message 60095.  

Hi guys,

I'm pretty familiar with BOINC but I'm unable to convince it to get multiple Milkyway@home tasks for my GPU. I'd love to do this as my internet connection is a bit dodgy at times and a single unit only runs for 2 and a bit minutes.

I've set the minimum and additional work buffer both to 1 day (but also tried with 2, 3 and 10 in either). I've got a couple of PrimeGrid tasks running on my CPU, but they've only got under an hour left each. There's no lack of drive space, my connection is all ok. All I get is the following message:

13476	Milkyway@Home	10/3/2013 7:30:24 PM	update requested by user	
13477	Milkyway@Home	10/3/2013 7:30:26 PM	Sending scheduler request: Requested by user.	
13478	Milkyway@Home	10/3/2013 7:30:26 PM	Not reporting or requesting tasks	
13479	Milkyway@Home	10/3/2013 7:30:27 PM	Scheduler request completed	


ANy idea what I can do to get some more tasks? Say 10 or 20 would be enough.


There is just not enough gpu units for everyone that wants them if we users cache that many, or in some case ALOT more, so the Project has a physical limit, but when you return a unit you will get a new one. I think the max is 4 per gpu per machine at any one time.
ID: 60103 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[DPC]Charley

Send message
Joined: 14 Feb 12
Posts: 4
Credit: 869,897
RAC: 0
Message 60109 - Posted: 5 Oct 2013, 7:22:20 UTC

Ok I get that, but I can't get 4 either. I can just get the one.

Also, it just doesn't fetch work quick enough. At times there's a whole minute (or more) when my GPU's simply idling because I can't get new WU's. Look at that in terms of WU's done: that's one WU not calculated for every two I do get :(
ID: 60109 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Mumak
Avatar

Send message
Joined: 8 Apr 13
Posts: 89
Credit: 517,085,245
RAC: 0
Message 60110 - Posted: 5 Oct 2013, 7:58:04 UTC

My Dual HD7950 machine has constantly 80 WUs in buffer.
ID: 60110 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,938,381
RAC: 22,848
Message 60112 - Posted: 5 Oct 2013, 12:46:51 UTC - in response to Message 60109.  

Ok I get that, but I can't get 4 either. I can just get the one.

Also, it just doesn't fetch work quick enough. At times there's a whole minute (or more) when my GPU's simply idling because I can't get new WU's. Look at that in terms of WU's done: that's one WU not calculated for every two I do get :(


Are you also doing PrimeGrid with your gpu? If so it could just be a cache setting or the very intricate 'debt' value that some projects still use.
ID: 60112 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[DPC]Charley

Send message
Joined: 14 Feb 12
Posts: 4
Credit: 869,897
RAC: 0
Message 60115 - Posted: 5 Oct 2013, 20:29:02 UTC - in response to Message 60112.  

Ok I get that, but I can't get 4 either. I can just get the one.

Also, it just doesn't fetch work quick enough. At times there's a whole minute (or more) when my GPU's simply idling because I can't get new WU's. Look at that in terms of WU's done: that's one WU not calculated for every two I do get :(


Are you also doing PrimeGrid with your gpu? If so it could just be a cache setting or the very intricate 'debt' value that some projects still use.


I have done so in the past, but am not currently doing so.
ID: 60115 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,938,381
RAC: 22,848
Message 60116 - Posted: 6 Oct 2013, 13:10:31 UTC - in response to Message 60115.  

Ok I get that, but I can't get 4 either. I can just get the one.

Also, it just doesn't fetch work quick enough. At times there's a whole minute (or more) when my GPU's simply idling because I can't get new WU's. Look at that in terms of WU's done: that's one WU not calculated for every two I do get :(


Are you also doing PrimeGrid with your gpu? If so it could just be a cache setting or the very intricate 'debt' value that some projects still use.


I have done so in the past, but am not currently doing so.


Then I am not sure what the problem is, sorry.
ID: 60116 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile dskagcommunity
Avatar

Send message
Joined: 26 Feb 11
Posts: 170
Credit: 205,557,553
RAC: 0
Message 60119 - Posted: 6 Oct 2013, 18:24:53 UTC

Do you use Ressourceshare of 0% ?
DSKAG Austria Research Team: http://www.research.dskag.at



ID: 60119 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[DPC]Charley

Send message
Joined: 14 Feb 12
Posts: 4
Credit: 869,897
RAC: 0
Message 60122 - Posted: 7 Oct 2013, 9:24:55 UTC - in response to Message 60119.  

Do you use Ressourceshare of 0% ?


Thanks! :) This was the culprit.
ID: 60122 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : Getting a (small) buffer of GPU units

©2024 Astroinformatics Group