Message boards :
Number crunching :
Optimising CPU Usage
Message board moderation
Author | Message |
---|---|
Send message Joined: 19 Aug 12 Posts: 4 Credit: 3,679,178 RAC: 0 |
I have a number of 1.76 Milkyway@home N-Body Simulation (mt) 8C tasks in the job queue The auto update downloads new tasks fine just I get 1.46 Milkyway@home Separation tasks one at a time. It could be more efficient if the job mailer server distributed these "1.46 Milkyway@home Separation" tasks in groups based on the available CPU threads, which for me is 8. When people are running multiple CPU projects then receiving 1, 2 4, 8 tasks or any other number is not important. This would just help out those solely running MilkyWay tasks. So there is no disadvantage in sending out the single cpu tasks in groups. Thanks |
Send message Joined: 8 May 09 Posts: 3321 Credit: 520,476,947 RAC: 26,152 |
I have a number of 1.76 Milkyway@home N-Body Simulation (mt) 8C tasks in the job queue This is an ongoing, as in more than a year, problem that the Admins don't know how to fix. If you setup a zero resource share 2nd gpu project then it will get a task while MW waits 10 minutes before they send you the next batch of tasks. |
Send message Joined: 19 Aug 12 Posts: 4 Credit: 3,679,178 RAC: 0 |
Well it seems that very soon after posting this tread I am now receiving plenty of the Separation tasks. So whatever happened my tasks 1C and 8C are now much more balanced. I had an idea someone tweaked the server but computers have a direction of their own sometimes so maybe the server just tweaked itself :) |
©2024 Astroinformatics Group