🚀 go-pugleaf

In article <434@srs.UUCP> matt@srs.UUCP (Matt Goheen) writes:
>Why is it that people just can't face facts.  The 80386 is just plain
>faster than a 68020 of the same clock speed for doing your typical
>"benchmarky" type things.  I have yet to see a benchmark say anything
>to the contrary.  Even Sun's latest "A RISC Tutorial" says that the 80386
>benchmarks faster than the 68030!  Now, I'm not saying that the 80386 is
>a great processor (or that I would ever buy a machine that had one in it),
>but let's face the facts.

I challenge you to produce any non-Intel publication which says that an
80386 is faster than a 68030 of the same clock speed.  It's simply not
true, even if you can find such a statement in print.  Don't forget to
include timings both with and without TI's 68030 cache extender chips
(these provide the '030 with one clock cycle access (same speed as the
on-chip caches or a register) to as much memory as you can afford!).

Whether an 80386 is faster than a 68020 depends on the systems in which
they are benchmarked, the operating systems, the languages and of course
the benchmarks.  The fastest CPU that BYTE benchmarked in their recent
4 month series (so far) was the DSI-780, which is a 16MHz 68020.  The
perfomance of the two CPU's is close enough that the relative difference
is less than the uncertainties inherent in benchmarking such
architecturally different CPU's (the uncertainties are significant).

--alan@pdn