>From direct experience, the faster amigas (say, 40 Mhz 68040) will dance
circles around some comparable 68k hardware, just because they have a
few more cycles left over to do real processing, because more of the
system is on custom hardware.  Remember, given identical processors,
an Amiga running a mac emulator is generally faster than a mac by a few
percentage points.  (Why?  Because the CPU isn't constantly being harassed
about video and mice.)

I'd say '020 68k hardware is fairly slow, but '040 68k hardware is
probably faster than any i386, and many i486, but slower than almost any
i586.

Please read that entire sentance before flaming me saying that the Pentium
is "an i386".

Likewise, I have a SPARC which is not signficantly faster than my 68020
sun3, and which I'm fairly sure is slower than a sun3x; it's a 4/110, and
it's the slowest SPARC ever released.  The SLC is still pretty pathetic,
and I'd expect even a medium-speed 80386 to beat it, and definitely it
loses to a 486.  To be fair, the faster sparcs are *much* faster than
most of the old CISC'y hardware.

Still, a lot of the actual machines we'll run on, say, from around the
time of the IPC/IPX/LX, are fairly weak.  I used an LX for about two
years, and my 68040 amiga was within 1% of it on every benchmark I tried.

I am told there are Alpha systems which are fairly horrible, too...

Still, overall, the table is in about the order I'd have put it in.  I
know nothing about the pmax.  I would probably have put the hp300 on the
same scale as all the other 68k, just because, from what I've seen, it's
about the same range of processors.