39 | | The maximum number of jobs sent to a given host in a 24-hour period is N times NCPUS. |
40 | | Helps prevent hosts with download or application problems from returning lots of error results. |
41 | | Be sure to set it large enough that a host does not go idle in a 24-hour period, |
42 | | and can download enough work to keep it busy if disconnected from the net for a few days. |
43 | | {{{ |
44 | | <daily_result_quota_cuda>N</daily_result_quota_cuda> |
45 | | }}} |
46 | | If present, the daily job quota is augmented by this quantity |
47 | | times the number of CUDA devices. |
| 39 | Each host has a field max_results_day in the interval [1 .. daily_result_quota]; |
| 40 | it's initially daily_result_quota, and is adjusted as the host sends good or bad results. |
| 41 | The maximum number of jobs sent to a given host in a 24-hour period is MRD times (NCPUs + NCUDA*cuda_multiplier). |
| 42 | |
| 43 | Set daily_result_quota large enough that a host can download enough work |
| 44 | to keep it busy if disconnected from the net for a few days. |
| 45 | {{{ |
| 46 | <cuda_multiplier>N</cuda_multiplier> |
| 47 | }}} |
| 48 | See the above. |
| 49 | Set this to the speed ratio of an average CUDA device to an average CPU. |