Message boards :
Number crunching :
N-body code - preference to limit # of CPUs?
Message board moderation
Author | Message |
---|---|
Send message Joined: 16 Aug 09 Posts: 1 Credit: 96,797 RAC: 0 |
I'm looking for a way to limit the max # of CPUs used by the milky-way N-body 1.5 application. It currently uses the Max CPUs in my boinc configuration, but that prevents any other CPU-based tasks from running while that's going on. Is there a separate config file I need to edit? Sat 24 Oct 2015 08:11:29 AM EDT | Milkyway@Home | Starting task ps_nbody_10_07_15_orphan_sim_1_1437561602_4262909_2 using milkyway_nbody version 152 (mt) in slot 4 |
Send message Joined: 8 May 09 Posts: 3339 Credit: 524,010,781 RAC: 0 |
I'm looking for a way to limit the max # of CPUs used by the milky-way N-body 1.5 application. Nope the N-body units are designed to do that, just exclude them and run the normal cpu units. |
Send message Joined: 22 Jun 13 Posts: 44 Credit: 64,258,609 RAC: 0 |
Ralph Hyre, There is a rather obscure option in the app_config.xml file to limit the number of CPU's used by the multi-thread N-Body tasks. Put together an app_conf.xml file that looks like the one below and store it in the Boinc\projects\milkyway.cs.rpi.edu_milkyway directory. The line that says <avg_ncpus>8</avg_ncpus> tells BOINC how many threads to reserve in it's job scheduling logic for the task. The line that says <cmdline>--nthreads 8</cmdline> tells the N-Body task how many threads to use. You will need to set those two lines to the number of threads that you want the N-Body tasks to use. <app_config> After you save the app_config_xml file in the milkyway folder, either restart BOINC or open the BOINC manager and tell it to "Read Config Files" (one of the options that you can get to from the menu bar at the top of BOINC Manager). In the Event Log, you should find a line that says "Milkyway@Home (data time) Found app_config.xml". Then download some N-body tasks and see if they conform to the new parameters. If you run into problems, you should be able to do a search for app_config.xml and find some useful hints. Hope that helps. |
Send message Joined: 8 May 09 Posts: 3339 Credit: 524,010,781 RAC: 0 |
Ralph Hyre, Huh live and learn...THANKS!! |
Send message Joined: 16 Dec 13 Posts: 23 Credit: 49,014,197 RAC: 4,222 |
Personally, I prefer the multi-threaded apps for the lower memory overhead and other benefits. If you're concerned about splitting up your contribution between multiple projects, the BOINC resource share is supposed to do that. Click here to see My Detailed BOINC Stats |
Send message Joined: 9 Jul 08 Posts: 85 Credit: 44,842,651 RAC: 0 |
This was exactly what I needed. I knew about <avg_ncpus> from other projects, but without <cmdline>--nthreads, n-body tasks were still behaving as *Must use ALL the THREADS* and choking off GPU utilization. Now I'm still able to use most cores for CPU work and still keep the GPU over 90% utilized. :-) Thanks captainjack. |
Send message Joined: 17 Feb 16 Posts: 14 Credit: 11,121,737 RAC: 0 |
I was looking for this as well. My newly modified app_config file is working great. <app_config> <app> <name>milkyway_nbody</name> <gpu_versions> <gpu_usage>.5</gpu_usage> <cpu_usage>.5</cpu_usage> </gpu_versions> </app> <app> <name>milkyway</name> <gpu_versions> <gpu_usage>.5</gpu_usage> <cpu_usage>.5</cpu_usage> </gpu_versions> </app> <app> <name>milkyway_separation__modified_fit</name> <gpu_versions> <gpu_usage>.5</gpu_usage> <cpu_usage>.5</cpu_usage> </gpu_versions> </app> <app> <name>milkyway_nbody</name> <max_concurrent>1</max_concurrent> </app> <app_version> <app_name>milkyway_nbody</app_name> <plan_class>mt</plan_class> <avg_ncpus>1</avg_ncpus> <cmdline>--nthreads 1</cmdline> </app_version> </app_config> Everything else runs on one CPU, N-Body should as well. 7 CPU tasks and 2 GPU tasks at a time keep everything loaded nicely. |
©2024 Astroinformatics Group