On Monday 16 April 2001  2:10 pm, Ben Harris wrote:
> On Wed, 11 Apr 2001, Chris Gilbert wrote:
> > See attached patch, (it's only for iomd based systems though, probably be
> > useable on the others though with a bit of tweaking)
> >
> > If anyone can give me some feedback on the size of the loop used to
> > calibrate that'd be good :)  I believe that it needs to be big otherwise
> > it might go too fast on the SA It may actually have to be based on the
> > cpu id though, or it may take too long on a the 600/700's
>
> It looks fine to me on both a 233 MHz SA-110 and a 40 MHz ARM610.  The
> ARM610 takes less than 100ms to do the calibration loop, so you can safely
> make it ten times bigger if you want (to allow for even faster CPUs).

Excellent, I might do that :)

> > If there are no objections I'll look to check the code in on friday.
>
> Oops.  Maybe I should have failed to object sooner.

Nope didn't put it in, someone pointed out that it should be more generic, 
and as all platforms support microtime, it seems to make sense to move the 
delay and delay calibration stuff into arch/arm/arm32 and call arch specific 
microtime :)

I plan to look at this tonight :)

Cheers,
Chris