Subject: Re: probing CPU speed?
To: Andrew Gillham <firstname.lastname@example.org>
From: Chris G. Demetriou <email@example.com>
Date: 11/19/1998 12:05:55
[ cc'd port-alpha, bcc'd port-i386; this is a port-alpha question,
take followups there. 8-]
Andrew Gillham <firstname.lastname@example.org> writes:
> On a slightly different note. Is is possible to adjust this HWRPB
> value? A friend of mine has a UDB with a dead battery, and it now
> thinks it is a 500Mhz Alpha! I don't know if it is a result of
> that counter or not, but it fails to drop to /bin/sh when I boot
The HWRPB is a software entity, created by the console software at
I don't know from where it gets the cycle counter frequency, but it's
either reported by the CPU or calculated. (Any ideas, ross?)
If the HWRPB has the wrong value, something Seriously Wrong is
happening, either at the hardware level or the console software level.
That may or may not be affected by configuration parameters stored in
non-volatile RAM, but i'm not aware of any alphas in which the CPU
speed is stored in non-volatile RAM.
The value given in the RPB as the cycle counter frequency is
indirectly used to calibrate delay() on the alpha, so if it's wrong,
you may run into all kinds of problems.
So, in a nutshell: I don't know what you can do about it, but
NetBSD/alpha's usage is correct and if the system is really reporting
that, it has a serious hardware and/or firmware problem.
Chris Demetriou - email@example.com - http://www.netbsd.org/People/Pages/cgd.html
Disclaimer: Not speaking for NetBSD, just expressing my own opinion.