Message boards :
Number crunching :
Nbody without disabling GPU?
Message board moderation
Author | Message |
---|---|
Send message Joined: 15 Dec 09 Posts: 12 Credit: 1,407,921,436 RAC: 0 |
I am running a 1700xt with 16 Gig PC=3300 with 2 Vega Frontier 16g. I can't run N-Body with out it disabling my GPU's. I will post my XML file in case i miseed simthing please look below and post if we can do this, Hope to get some help with my problem, thanks pp_config> <app> <name>milkyway</name> <gpu_versions> <gpu_usage>0.25</gpu_usage> <cpu_usage>0.25</cpu_usage> </gpu_versions> </app> <app_config> <app> <name>milkyway_nbody</name> <max_concurrent>1</max_concurrent> <app_version> <app_name>milkyway_nbody</app_name> <plan_class>mt</plan_class> <avg_ncpus>12</avg_ncpus> </app_version> </app> </app_config> |
Send message Joined: 15 Dec 09 Posts: 12 Credit: 1,407,921,436 RAC: 0 |
sorry missed a bracket in copy past. |
Send message Joined: 27 Jul 14 Posts: 23 Credit: 921,261,826 RAC: 0 |
First, check your startup log. There should be a message saying your app_config is read, or a message saying there is a problem. It does look like there's a syntax problem. <app_config> is in there twice. <app_config> <app> <name>milkyway</name> <gpu_versions> <gpu_usage>0.25</gpu_usage> <cpu_usage>0.25</cpu_usage> </gpu_versions> </app> <app> <name>milkyway_nbody</name> <max_concurrent>1</max_concurrent> <app_version> <app_name>milkyway_nbody</app_name> <plan_class>mt</plan_class> <avg_ncpus>12</avg_ncpus> </app_version> </app> </app_config> I don't run nbody but see if that works. You might need to use the cmdline parameter to control the number of nbody threads instead of ncpus. https://boinc.berkeley.edu/wiki/Client_configuration Team USA forum | Team USA page Always crunching / Always recruiting |
Send message Joined: 15 Dec 09 Posts: 12 Credit: 1,407,921,436 RAC: 0 |
Thanks, will look into the CPU commands in the link and will give it a try. The two clients, one for GPU and CPU sounds less of a pain if i can figure it out, and if not I'll be back. |
Send message Joined: 24 Jan 11 Posts: 708 Credit: 544,691,669 RAC: 113,173 |
On the face of it, seems you don't have enough cores to satisfy both the gpus and the cpu. Your mt application wants to use 12 cores out your total of 16 available for your cpu. Then you want to run 4 gpu tasks per card and you have two cards. So a total of 8 gpu tasks running. Even with only .25 cpu cores to support each task, that adds another 2 cores to your 12 mt cores leaving only 2 cores to support the desktop and background tasks. And the <cpu_usage>0.25</cpu_usage> is only a scheduling recommendation. The gpu tasks will use as much of a cpu core as the science application needs. It can be a lot more than just a quarter of a core. If you have BoincTasks running it will report that actual amount of cpu usage a gpu task required to run it. I run my Einstein gpu tasks with a full cpu to support it so I don't have any typical values about what a task actually needs minimally to support it. Maybe others can jump in with typical values. |
Send message Joined: 24 Jan 11 Posts: 708 Credit: 544,691,669 RAC: 113,173 |
I would just drop your mt cpu core usage down to eight and see if both task types run. |
©2024 Astroinformatics Group