Message boards : Questions and problems : Offloading inactive clients
Message board moderation
Author | Message |
---|---|
![]() Send message Joined: 29 May 09 Posts: 26 ![]() |
I'm involved with seti@home, milkyway@home and cosmology@home. Once these clients become resident in RAM they stay there, even though BOINC has switched to another project. The cosmology client, camb_2.16_windows_intelx86.exe takes up 335 MB, so it is the worst offender, but astropulse_5.03_windows_intelx86.exe uses 50 MB, so it would be nice to free up the memory. These numbers include the data files the clients are working on. Is there any way to get BOINC to offload these files from RAM when they are no longer in use? |
![]() Send message Joined: 8 Jan 06 Posts: 448 ![]() |
I'm involved with seti@home, milkyway@home and cosmology@home. Once these clients become resident in RAM they stay there, even though BOINC has switched to another project. The cosmology client, camb_2.16_windows_intelx86.exe takes up 335 MB, so it is the worst offender, but astropulse_5.03_windows_intelx86.exe uses 50 MB, so it would be nice to free up the memory. These numbers include the data files the clients are working on. Is there any way to get BOINC to offload these files from RAM when they are no longer in use? The computing preference in the project has an option. Leave applications in memory while suspended? (suspended applications will consume swap space if 'yes') You will notice that it says 'consume swap space' that means that it isn't necessarily using RAM but the OS will move it temporarily to a swap file on the hard drive if it needs the RAM for something else. That's the trick the OS uses to run many programs at the same time on a limited amount of memory. PS If you remove the application completly, suspended WU may have to restart at an earlier point than it left off. Boinc V 7.4.36 Win7 i5 3.33G 4GB NVidia 470 |
![]() Send message Joined: 20 Dec 07 Posts: 1069 ![]() |
PS If you remove the application completly, suspended WU may have to restart at an earlier point than it left off. That's not the case for SETI and Einstein, i.e. no loss of work done. |
![]() Send message Joined: 29 May 09 Posts: 26 ![]() |
Thanks. That's just what I needed. |
![]() Send message Joined: 8 Jan 06 Posts: 448 ![]() |
There I thought the newer clients switched at checkpoint time, not to loose progress so LAIM is not really needed, but for a slower resume. Then if LAIM is not on and your jobs are pre-empted due for instance high priority jobs and periodic benchmarking, they do get unloaded. That SETI and Einstein are 'forced' to write a resume point is interesting in that respect. My understanding is fairly sketchy on how the OS swaps to/from VM are performed, but what you see is probably normal. I should think it only swaps out blocks of memory as needed, not the entire apps. |
Copyright © 2025 University of California.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License,
Version 1.2 or any later version published by the Free Software Foundation.