Re: Linux gets O(1) scheduler
Martin Husemann <[email protected]>
| Newsgroups | gmane.os.netbsd.devel.performance |
|---|---|
| Message-ID | <[email protected]> |
On Sat, Dec 27, 2003 at 01:51:39AM +0100, Jesper Louis Andersen wrote: > Ergo, we can select the next proces in O(c) = O(1) time. Yeah, sure. The interesting question is how much work needs to be done in advance to manage the different queues. And as you sketched it, the alternative array. Big oh is nice in the academic world. But at the scheduler, the details matter. So what would be far more interesting than a pointer to a news article bringing this up in marketing speech to the masses, would be a pointer to a paper describing the details. Martin