Advanced search

Message boards : Wish list : Max # of WU's

Author Message
Profile Laura_Stevens
Avatar
Send message
Joined: 30 Sep 08
Posts: 6
Credit: 50,092,565
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 8790 - Posted: 23 Apr 2009 | 14:38:25 UTC

I think it would be helpful, and both in the best interest of the project as well as the crunchers, if in the GPUGRID preferences there was an option for how many WU's should be allowed to download at a time. I'm not sure how feasible it is, but it could be quite helpful.

I have noticed that often, with a quad core CPU and 1 GPU, it will try to maintain 4 tasks, which leads to a longer turn around time, which leads to many of the tasks being redundant. If there were an option to set for example, MAX_WU's = 2, then it would improve turn around time for those that can take advantage of it. I know it's not going to help some users, but I think it could be good for many. Note: option would be for concurrent tasks, not daily tasks.

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 8795 - Posted: 23 Apr 2009 | 17:55:20 UTC - in response to Message 8790.

I think it is all in the size of the "Maintain enough work for an additional..." setting in your computing prefrences. If you set this to be just under 1 day you should end up with the right number of WUs for your GPU to return within 24 hours which has the side benefit of reducing the number of "cancelled redundants".

Steve

Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 8796 - Posted: 23 Apr 2009 | 19:12:55 UTC - in response to Message 8790.

Try setting the work buffer in BAM preferences on the PC in question to 0.2, that should keep you topped up without excessive downloads.

Regards
Zy

Profile Dan T. Morris
Send message
Joined: 19 Jun 08
Posts: 13
Credit: 806,868
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 8808 - Posted: 23 Apr 2009 | 23:12:27 UTC

It would be my wish that the project powers that be grant larger credit for wu. With Milky way going to Nvidia GPU you may see a large drop in users, also please add support for the ATI HD series cards. They are much faster and cheaper than Nvidia.

Thank You.

DD

Profile Laura_Stevens
Avatar
Send message
Joined: 30 Sep 08
Posts: 6
Credit: 50,092,565
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 9298 - Posted: 4 May 2009 | 13:41:34 UTC
Last modified: 4 May 2009 | 13:42:07 UTC

It seems after upgrading to a newer version of BOINC, the tasks are now properly obeying the cache settings. Apparently, the problem was with BOINC 6.4.2, it would grab 1 WU for each core of the CPU, even if cache was set to 0

TomaszPawel
Send message
Joined: 18 Aug 08
Posts: 121
Credit: 59,836,411
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 9672 - Posted: 12 May 2009 | 12:57:03 UTC - in response to Message 9298.

I also have Q8200 and 8600GT, and it downloads 4 WU....

I want option how many WU will be download...
____________
POLISH NATIONAL TEAM - Join! Crunch! Win!

Furlozza
Send message
Joined: 30 Jan 09
Posts: 7
Credit: 100,131,811
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 11539 - Posted: 31 Jul 2009 | 11:49:15 UTC - in response to Message 9672.

Under the new system, I should carry 4 WUs at a time on an E6850 plus GTS250, meaning that there is a good chance that two of the jobs will be a waste of time and although finishing will go unrewarded.

Can it not be done within GPUGrid prefs in some way or manner that the cruncher actually has a direct say in how many units are downloaded/maintained.... or could the info about puters and GPUs that are already available to you be actually used as a guide to WUs to be downloaded?

John
Send message
Joined: 27 Jun 09
Posts: 3
Credit: 31,290,058
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwat
Message 11817 - Posted: 12 Aug 2009 | 2:25:39 UTC

I would like to see either more WU's allowed or not to consider an uploading WU as an outstanding WU.
Right now, I have one unit uploading very slowly and another in progress. As has happened before, my GPU will run out of work before I receive another unit.

John
Send message
Joined: 27 Jun 09
Posts: 3
Credit: 31,290,058
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwat
Message 11878 - Posted: 14 Aug 2009 | 3:06:48 UTC - in response to Message 11817.
Last modified: 14 Aug 2009 | 3:51:12 UTC

I'm replyng to my own post.
My GPU is now out of work. I now have two WU's completed and uploading very slowly, but can't get another WU, because I'm limited to two.
I'm looking for another project to keep keep my GPU busy, but the aqua@home has been giving me "Computation Error". Trying to work thru that.
Anyone know of another good project that I can attach to that could make use of my GPU?
OK, never mind, I've attached to seti@home now.
____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 11899 - Posted: 14 Aug 2009 | 9:42:00 UTC - in response to Message 11878.

With two wus in progress you should never get out of work.
Increase the resource share for gpugrid or something similar (moderators can tell you more) will make sure that you always request enough WUs.

gdf

Post to thread

Message boards : Wish list : Max # of WU's

//