Re: Process scheduling.

Greg Hewgill (
Mon, 9 Jun 1997 00:14:51 -0700

> From: Colin L. Hildinger <>
> I like OS/2's method, personally. There are four brackets of priority,
> each with higher importance. A critical priority always takes
> precedence over an idle one. Period. [...]
> BTW - how does it work in NT? I use it but I have to admit I'm not
> intimately familiar with the scheduling.

NT has pretty much exactly the same scheduling algorithm as OS/2 (a
surprising amount of the base architecture of the two systems is the same
actually). When running an idle priority process, anybody that has any "real"
work to do gets the CPU. I understand that occasionally NT will juggle
process priorities temporarily to avoid total starvation, but this is not
noticeable on any performance meter. I would suspect that OS/2 may do the
same thing.

> Same question for 95 I guess,
> but whatever 95 does, it's poorly implemented based on the useability
> of the system when you get a few processes going, especially DOS
> programs. 95 can't deal with them to save its neck.

The scheduler is Windows 95 is much different, and definitely worse for
background idle tasks like DESChall. DOS programs really test the scheduler,
as they generally don't voluntarily give up any CPU at all. Just Say No! :)