Please can we not use clock speed of a measure of worth of CPUs. It
doesn't mean much. Equally well I oculd argue that SPARC boxen are
better because they are quad issue and not double as many PCs or the
fact that they are triple address not double...
I haven't done this, this is why I mentioned the parallelism of
the cpu in my equations.
And most PCs are not dual issue, they are quad issue just like
the UltraSPARC family.
Well that depends on whether you are talking about x86 code or IOPs and
also which particular x86 CPU you are talking about. Then again we
could get on to talking about pipelineing, branch prediction, memory
latency, relative cache sizes (I may be using a 333Mhz UIIi but it has a
2Mb cache - which feels like it makes quite a difference). The point I
was trying to make is that there aren't any easy numbers you can compair
between CPUs to say 'this one is better' - comparision is vital but is
also extreemly difficult.