Re: [f-cpu] some technical stuff (yeah :-D)

[Yann Guidon <whygee@f-cpu.org>]

hi !

Michael Riepe wrote:
> On Fri, Jun 15, 2001 at 01:45:55AM +0200, Yann Guidon wrote:
> [...]
> > > What about a variable internal clock?  Using a (programmable) NCO,
> > > we could increase/decrease the clock frequency without any glitches.
> > that's what i consider most.
> > On top of that, the clock can be controlled by a HW real time controller
> > which analyses how many instructions are executed etc...
> > so the NCO can be matched to the real CPU activity.
> 
> I'd rather add another special register that allows the CPU to change the
> NCO frequency.  Analysing usage patterns, detecting idle periods and so
> on is the operating system's task.

aziouwiche :-)

> > i'm pretty sure that somebody has already patented it bu why care ?
> 
> I don't know.  Several CPU vendors use programmable clocks (combined
> with variable core supply voltages) in their low-power/"mobile" CPUs.
> I don't know *how* they do it, but an NCO is the obvious way, IMHO.
> And it's rather easy to implement (two registers and an adder).  The hard
> part is to keep the F-bus and the SDRAM-interface going at the same
> speed all the time.

i don't think it's the major trouble.
the speed difference between the core and the memory is already so large
that the core is designed to take that into account. Varying the clock speed
will not change/jeopardize the FC0. There can be buggy implementations
but i am not responsible :-)

>  Michael "Tired" Riepe <Michael.Riepe@stud.uni-hannover.de>
WHYGEE
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*************************************************************
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe f-cpu       in the body. http://f-cpu.seul.org/