henry@utzoo.UUCP (Henry Spencer) (01/31/84)
Some skeptical comments on Robert Moore's observations:
The 68020 will contain a cache for text refrerences, with
only 2 clock cycles per instruction running from the
cache. The 16k doesnt have a cache.
I hope this "cache" is going to be bigger than the two-word one on
the 68010, or else it's useful only for "loop mode".
The 68020 provides ...
... memory resident pointer addressing modes.
What's that? And what good is it? Have you checked out the 16k's
"external" addressing mode?
The 68020 fpu, due out 3 months after the 020, provides ...
[lotsa good stuff]
This does sound good, assuming it runs at decent speed. National's
FPU is not bad but is not a world-beater. But "3 months"??? Come
off it. Especially since it's relative to the arrival of the 68020,
which is itself off in the never-never. (Especially if one wants a
bug-free version.)
The 68020 will be provided with clocks from 12 to 25 Mhz. The
design center is around 20 Mhz. The 16k is hard to get
above 6 Mhz. They are hopeing for 10 Mhz eventually.
They claim to have 10 MHz right now. I haven't actually bought the
parts to check how well they work, but this sounds better than a claim
that the "design center" is around 20 MHz. When do we actually get to
see parts that *run* at 20 MHz? If it's at the same blinding speed that
we got to see 10 MHz 68000's, I'm not holding my breath. Note that the
folks at National have been heard mumbling about higher clock speeds for
the 16k, once they have the 10 MHz stuff rock solid. Any bets on who
gets there first?
The 68020 paged mmu will allow variable page size and number of
levels. This is fixed in the 16k's mmu.
If the 68020 "paged mmu" is going to be a clone of the existing 68000
mmu, I don't want to even see a spec sheet. I no longer have any faith
in Motorola's ability to get things right in an mmu. I see no great
use for variable number of levels, and "variable page size" is dead easy
to get on the 16k, assuming you're willing to make your page sizes
multiples of 512. I trust Motorola will insist that they be powers of
2, as before?
These and other considerations make the 68020 worth waiting for.
Not to me. Ok, so I'm biased. I still think that Motorola is making
an awful lot of promises about the 68020 and may take quite a while to
make good on them.
--
Henry Spencer @ U of Toronto Zoology
{allegra,ihnp4,linus,decvax}!utzoo!henrymats@dual.UUCP (Mats Wichmann) (02/15/84)
Ah, yes, the long awaited 68020 and the even more mysterious
MMU chip to go with it. For those of you who are worried about the
MMU, they got the message and threw out the 68451 design. In
fact, I understand that Motorola stole away some of the guys that
did the National MMU, so they should be somewhat similar. Ever wonder
why the top people always seem to be leaving National??? Or do they
just get bad publicity? We have all of the most current poop on the 68020,
all stamped Motorola Company Confidential, so I can't really elaborate
on the exact details, although nobody has been far off the mark.
The MMU is a different question entirely - details are sketchy,
primarily because even the paper design is not complete yet.
3 months after the 020 may be possible - only if the 020 is as late
as some of their other chips have been. If you are trying to
design a product using the trio - CPU, MMU, and FFP, you will
be waiting a long time before that product comes to market.
But is the 020 worth waiting for? Depends on your perspective -
from MY jaded point of view, neither the 020 OR the 32032 is
particularly new or interesting techology, and nothing else
really new seems to be turning up elsewhere...so it depends
on other factors - when do you need a machine, for how much,
what should it do, and so forth. Is it worth waiting a year
for a system running the 020 when you could have spent that
time in productive work on a 68000? In general, I don't think
the chip inside the system makes that much difference. What
really makes the difference is how the machine runs your
application, how reliable it is, and how well it is supported.
And THIS really comes down to the qualtiy of the vendor and what
he has done with his product. Okay, enough with the soapbox...
Mats Wichmann
Dual Systems
Berkeley, Ca.
{ucbvax,amd70,ihnp4,cbosgd,decwrl,fortune}!dual!mats