| 1 | = Non-CPU-intensive applications = |
| 2 | |
| 3 | A 'non-CPU-intensive' project is one whose applications use little CPU time. Examples include: |
| 4 | * Host measurements |
| 5 | * Network measurements |
| 6 | * Web crawling |
| 7 | * Network data access |
| 8 | |
| 9 | A non-CPU-intensive project is treated specially by the core client: |
| 10 | * The core client maintains one result for the project |
| 11 | * The core client executes this result whenever computation is enabled, bypassing the normal CPU scheduling mechanism. |
| 12 | |
| 13 | A project labels itself as non-CPU-intensive by including the <non_cpu_intensive> flag in its [http://boinc.berkeley.edu/configuration.php configuration file] |
| 14 | |
| 15 | This attribute is at the project level; there is no provision for a project to have some applications that are CPU intensive and some that aren't. |
| 16 | |
| 17 | Non-CPU-intensive applications can use chunks CPU time; this won't break anything, and the CPU scheduler will adjust to it by changing its estimates of 'CPU efficiency'. However, non-CPU-intensive applications should try not to use more CPU time than their resource share fraction indicates. |
| 18 | |