| 1 | = Client app configuration = |
| 2 | |
| 3 | This doc describes a new feature that will allow volunteers to adjust |
| 4 | application scheduling parameters. |
| 5 | |
| 6 | Note: this is separate from the anonymous platform mechanism. |
| 7 | |
| 8 | A volunteer can configure a project's apps by putting a file '''app_config.xml''' |
| 9 | in the project directory. |
| 10 | This file has the following form: |
| 11 | {{{ |
| 12 | <app_config> |
| 13 | <app>name</app> |
| 14 | <max_instances>N</max_instances> |
| 15 | <gpu_versions> |
| 16 | <gpu_usage>.25</gpu_usage> |
| 17 | <cpu_usage>.1</cpu_usage> |
| 18 | </gpu_version> |
| 19 | </app> |
| 20 | </app_config> |
| 21 | }}} |
| 22 | '''<app>''': the short name of an application. |
| 23 | |
| 24 | '''<max_instances>''': a limit on the number of concurrent jobs for that application. |
| 25 | |
| 26 | '''<gpu_versions>''': the '''<gpu_usage>''' and '''<cpu_usage>''' elements specify the |
| 27 | number of device instances used by GPU app versions. |
| 28 | Note: there is no provision for specifying this per GPU type or per device. |