Re: [OT?] MHz vs. MIPS in processors

From: Bob Myers (nospamplease_at_address.invalid)
Date: 06/30/04 

Date: Wed, 30 Jun 2004 04:13:55 GMT

"Chaos Master" <chaosmater@pop.com.br.INVALID> wrote in message
news:MPG.1b4adb3f84b209e7989bad@130.133.1.4...

> > : Q: For someone new in the area, what is the biggest difference between
telling
> > : processor speeds in MIPS and in MHz?
> >
> > The biggest difference is that MHz is a real quantifiable value of clock
> > speed, MIPS is a highly subjective marketing fluff word of little value.
>
> I always thought that MIPS = MHz or something like, so I sometimes said
that a
> 1GHz Pentium III processor was 1000 MIPS. Now things make more sense :P

Only if the processor in question executes, on average, one
instruction per clock cycle. That's not the norm, and in fact there
is no fixed number of clock cycles required per instruction that
applies to all processor architectures - so there's no equating
MIPS and MHz.

Having said that, though, the notion that "MHz is a real quantifiable
value of clock speed," as if that alone told you something about the
relative performance of various processors, is equally nonsensical.
You can compare performance by looking at clock speeds alone
ONLY if the processors in question have basically identical
architectures - but not in general. For instance, suppose we have
one processor which on average requires four clock cycles to
complete an instruction, while another requires only two. Further
assuming that the instruction sets of the two are comparable (in
simple terms, you "get as much done" per instruction, on average,
between the two), then the second processor at 2 GHz will outperform
the first at 3 GHz. So...it ain't that simple. And you definitely don't
have anything to brag about if you look at clock rates alone.

This whole situation is what leads to performance benchmark
testing, in which programs are written to simulate the sort of real-
world tasks that the processor will be called upon to do. You
run these benchmarks on the processors you want to compare,
and the one which finishes the job first is the higher-performance
processor. Simple, and not as subject to marketing spin as the
more-simplistic MIPS or MHz numbers alone.

Bob M.

Relevant Pages

  • Re: Why is SUN falling so far behind IBM?
    ... The CPU clock is quite irrelevant. ... bottlenecks while SGI works on resolving those instead. ... I would put forth that it's reasons like this why customers like ... > And that's where MIPS simply is behind all competition... ...
    (comp.unix.aix)
  • Re: Why is SUN falling so far behind IBM?
    ... Of course when comparing performance/clock cycle MIPS still ... The CPU clock is quite irrelevant. ... Yes, because SGI sold most of it's assets like Cray, Softimage, and also ... MIPS/IRIX sales are almost all just upgrade sales to the remaining existing ...
    (comp.unix.aix)
  • Re: microcontroller selection
    ... Strange criteria indeed. ... Both data width and clock speed are usually ... development system cost, peripheral requirements, etc., etc. ... a core inside an FPGA) can run from a 12 MHz clock, some just under 1 MIPS ...
    (comp.arch.embedded)
  • Re: non load/store architecture?
    ... But the ColdFire code is more compact, ... instruction, giving better performance for the same ipc. ... Only data cache. ... and neither are most fast RISC cores as they need high clock ...
    (comp.arch.embedded)
  • Re: one way light speed
    ... You can use one clock to measure the two-way speed of light ... You still need to decide on how to synchronise your clocks. ... is accepted by all that the two-way speed of light is c in all directions. ... Thus comparing the speeds. ...
    (sci.physics.relativity)