Re: [announce] [patch] ultra-scalable O(1) SMP and UP scheduler

[Davide Libenzi]

On Sun, 6 Jan 2002, Ingo Molnar wrote:

> And George Anzinger has a nice idea to help those platforms which have
> slow bitsearch functions, we can keep a floating pointer of the highest
> priority queue which can be made NULL if the last task from a priority
> level was used up or can be increased if a higher priority task is added,
> this pointer will be correct in most of the time, and we can fall back to
> the bitsearch if it's NULL.

Ingo, you don't need that many queues, 32 are more than sufficent.
If you look at the distribution you'll see that it matters ( for
interactive feel ) only the very first ( top ) queues, while lower ones
can very easily tollerate a FIFO pickup w/out bad feelings.

- Davide