787 |
|
* apply transiently, so cheapen this by (a) not bothering to |
788 |
|
* weight at all unless there is more than one queued task (b) |
789 |
|
* rather than compensating by a factor of (#idle/#active) |
790 |
< |
* threads, we just substract out a function of #idle that is |
790 |
> |
* threads, we just subtract out a function of #idle that is |
791 |
|
* a good enough approximation for conditions near the |
792 |
|
* borderlines for threshold testing. This errs in the |
793 |
|
* direction of reporting more extreme lack of surplus (as in |