Re: When will the YDL 2.2 iso be available for FTP download?


Subject: Re: When will the YDL 2.2 iso be available for FTP download?
From: Timothy A. Seufert (tas@mindspring.com)
Date: Tue Mar 19 2002 - 21:30:36 MST


At 1:35 PM -0600 3/19/02, Robert Brandtjen wrote:
>On Monday 18 March 2002 11:49 pm, Pat Plummer wrote:
>> Yeah, well, my emotional bias is toward the PPC chip and this is a PPC
>> based distro/list :). Sorry ...
>
>I too once believed in the power of the PPC platform -only to see it's early
>triumphs over intel slowly undermined to the point where it could barley hold
>it's head up - sorry, but this is just too true. I believe it is mostly due
>to a Jobs-IBM fallout - Motorola is not IBM, IBM demoed a working 1GHZ G3
>chip at macworld in 1998 - whatever happened to it?

I shouldn't get into this one too, but what the heck, this is such a
common myth, and it's hardly your fault that you've fallen prey to
it, since it's been reported as fact so often...

There was a real demo which spawned the myth. It was a PPC-like
integer only thing incapable of booting any real OS -- not a complete
processor by any stretch of the imagination. IBM was using it to
demonstrate the progress they'd made in designing and fabricating GHz
digital logic. A tech demo, not a product demo.

Unfortunately some elements of the Mac press picked up on it because
IBM mentioned "PowerPC" in the press release. The game of Telephone
was played, and suddenly it became a GHz G3 waiting in the wings.
Made a better story that way I guess.

Having the demo take place at MacWorld Expo is a mutation of the
story I hadn't heard before. (It really happened at a semiconductor
industry conference.)

>It took motorola nearly 4
>years to reach that speed with the G4.

There have been two problems. One is that Motorola had problems with
developing new chip fabrication technologies, and therefore could not
process shrink the G4 for a long time. The other is that the
original G4 core design has roughly the same speed limitations in any
given process as the G3, since it's basically the same thing plus
AltiVec. IBM has been able to push the G3 to clock frequencies
higher than 500 MHz mainly through a series of process shrinks, with
only the very latest 0.13 micron shrink claimed to even be capable of
scaling to 1 GHz.

Motorola now is merely 1 process generation behind IBM (they're at
0.18 micron), but they have a new core design capable of higher
frequencies, the 7450 family (AKA G4+). Every Mac with a G4
processor 550 MHz or faster uses a 7450 derivative. Despite
Motorola's continuing process tech lag, the 7450 has consistently
achieved better clock rates than IBM's shrunk 750s ever since
introduction of the 7450.

As for the paranoid conspiracy stuff about Apple keeping Briqs from
getting faster CPUs -- traditionally Apple gets the lion's share of
production when Motorola is having a hard time producing enough
processors to meet overall demands, but otherwise Motorola will sell
to anybody with cash. Right now, at a minimum, 733 MHz 7450
processors should be available to anybody who wants them, that being
a speed bin Apple isn't even shipping any more.

Macs probably use less than 50% of all PowerPCs made -- possibly even
of the models that Apple uses in Macs. PowerPC is very popular in
embedded computing, including companies ranging in size from Total
Impact (makers of the Briq) to Cisco (which is using the 7450 in some
kind of router product, I believe). Motorola isn't going to cut off
that kind of revenue just for Apple's sake.

What it really comes down to is re-engineering. A 1 GHz 7450 uses *4
times* as much power as a 500 MHz 7410. The existing Briq has a
pretty simple cooling system that is probably inadquate. Also,
they'd need to redo the motherboard as the 7450 is not pin compatible
with the 7400/7410, nor is it available in the ZIF modules they use
in the current Briq. The 7450 simply isn't a drop-in replacement for
older CPUs.

The Mac upgrade companies must overcome that same heat problem, in
addition to power supply problems, bus voltage level differences, and
possibly bus protocol incompatabilities with older Mac bridge chips.
I have experience designing PowerPC embedded computers, I've looked
into the issues on a casual basis (out of curiosity), and these are
real engineering problems which may make it very difficult to pull
off 7450 upgrades for any system design older than Apple's Sawtooth
(G4 AGP) motherboard.

-- 
Tim Seufert



This archive was generated by hypermail 2a24 : Tue Mar 19 2002 - 21:45:23 MST