Subject: Re: fast timeouts (< 1 millisecond)??
To: Chris G Demetriou <Chris_G_Demetriou@ux2.sp.cs.cmu.edu>
From: Perry E. Metzger <firstname.lastname@example.org>
Date: 11/14/1996 21:25:16
Chris G Demetriou writes:
> > It would really be helpful to my driver if there was a mechanism to
> > schedule quick timeouts (ie. much less than 1/hz and usually on the
> > order of 10s of microseconds)?
> > Any comments?
> Well, i suppose i've got a couple of questions, like:
> (1) what would you have the system doing?
> (2) what method would you suggest for having the system figure
> out when to stop delaying?
Its clear that Hz will be rising as time goes on -- systems like the
Alpha already (I believe) have timer interrupts going off on the order
of every ms. Also, some machines might have spare counter chip slots
that can trigger interrupts.
How about we provide a kernel call to do fine granularity delay, and
on machines where there is some method (like a spare counter interrupt
or a very fast hz) to do it "right" it gets done right, and on
machines where there is no way we just spin?
PS I wonder what the impact of jacking up Hz on port i386 on faster
processors would be. A 200Mhz Pentium Pro clearly could handle ten
times the clock interrupts of some 33Mhz 486 or what have you...