Message boards : Number crunching : One hour crunchtime selected , even getting longer and longer, i am sad, possibly angry about that.
Author | Message |
---|---|
Emerald42 Send message Joined: 9 Jun 08 Posts: 9 Credit: 1,675,720 RAC: 0 |
Cheers, it´s great to crunch but i hate to wait to long to crunch and after it to see thats the runtimes of the wus are 200-300% longer as predicted. Thats for a long time of crunching and getting me upset. Now´s there a possiblity to get this thing running for only ONE hour per wu, or not ? Waiting for better times... Thx, reading. |
Mod.Sense Volunteer moderator Send message Joined: 22 Aug 06 Posts: 4018 Credit: 0 RAC: 0 |
One hour preferred runtimes will often see tasks running longer than that. How close to the one hour preference will vary by the type of work unit. So, depending on what your reasons are for wanting only one hour runtimes, R@h may not be an appropriate choice for you. Selecting a two or three hour preference should see much more consistent runtimes. Rosetta Moderator: Mod.Sense |
Montreal Crew Send message Joined: 9 Jan 18 Posts: 1 Credit: 153,170 RAC: 0 |
Rosetta work units run a number of simulation 'models' - each model might take 1 to 2 hours to run. There is no use in running 'half of a model' so what happens is if someone has a target run time of 6 hours, their Rosetta client will attempt to complete as many models as possible within 6 hours but the last model might take longer to finish and drive the total runtime slightly past 6 hours. |
rjs5 Send message Joined: 22 Nov 10 Posts: 273 Credit: 22,962,883 RAC: 17,460 |
You might check the amount of disk traffic you are seeing. Rosetta will/can take a good chunk of memory (200mb-400mb routinely and much more on occasion) and you might be exhausting your physical memory and paging/swapping memory to disk. If that is happening, you will see a much longer and variable run time. Your Rosetta jobs should take very little OS time. |
Message boards :
Number crunching :
One hour crunchtime selected , even getting longer and longer, i am sad, possibly angry about that.
©2024 University of Washington
https://www.bakerlab.org