> In the 2.0.27 and 2.0.29 kernels (RedHat distributed kernels with the
> cyrix.pl2.patch file applied for the Cyrix CPU features) have given me a
> BogoMIPS of 54.2 average while a 2.0.30 kernel that has only the cyrix
> patch applied gets 99.2 BogoMIPS.
>
> I am confused on this as these three kernels are compiled exactly the same
> would produce difference BogoMIPS for the same configuration options
It sounds like branch prediction then. The Cyrix patches fiddle with
the branch prediction settings - different patches do different
things, sometimes as a compile option. The bogomips value comes
from the calibration of the delay loop. This is very small and simple
and loops entirely in the processor's L1 cache. With branch prediction,
pipelining, speculative execution, register renaming and other bells
and whistles the timing loop can make the bogomips value *completely*
bogus (as opposed to merely misleading :-) ).
Mike
-- .----------------------------------------------------------------------. | Mike Jagdis | Internet: mailto:mike@roan.co.uk | | Roan Technology Ltd. | | | 54A Peach Street, Wokingham | Telephone: +44 118 989 0403 | | RG40 1XG, ENGLAND | Fax: +44 118 989 1195 | `----------------------------------------------------------------------'