NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/bit/listserv/iamiga/10571

Text File | 1993-01-23 | 10.6 KB | 225 lines

Comments: Gated by [e-mail]
Path: sparky!uunet!paladin.american.edu!auvm!!WINEBLOOD"
X-Envelope-to: [e-mail]
Content-transfer-encoding: 7BIT
X-Vmsmail-To: SMTP%"[e-mail]"
Message-ID: <[e-mail]>
Newsgroups: bit.listserv.i-amiga
Date:         Fri, 22 Jan 1993 12:16:35 -0700
Sender:       Info-Amiga List <[e-mail]>
From:         "Tech/Lab Manager In Training: Deva WinbloodDayva Wineblood"
              <[e-mail]>
Subject:      Info on PowerPCs, Motorola processors, and future speculations.
Lines: 210

This was sent to me by a friend, and I am not presently subscribed to
I-AMIGA, but I thought that you may wish to receive the info.

        Deva Winblood(Dayva Wineblood)  [e-mail]

------

Date: 20 Jan 93 16:14:40 GMT
From: [e-mail] (Noah Price)
Subject: Motorola's 68060 for the Mac

In article <mhk=[e-mail]>, [e-mail] (Chung Hsiung
Yang) wrote:
>       I thought that the 68060 was supposed to compete with the
> INTEL P5 chip.  If so the 68060 could be a 64 bit chip.  I may be
> wrong of course.

I recently got the Fall 92 issue of a little newsletter called "68K
Connection."  It says Vol 1 Issue 1, so I guess it's the first.  In any
case, they had a table summarizing the M68000 family.  It had columns for
68000, 68020, 68030, 68040, and 68060.

In that table, they listed the following info about the 060.  I'm sure this
has all been published before, but here it is straight from a public Moto
brochure.

MIPS               100+                       (vs. 39 for the 040)
MFLOPS             12                        (vs. 3.5 for the 040)
Address Range      4G Byte
Data Bus           32 bit
Clock Speeds       50-66 MHz         (040 was listed as 25-40 MHz)
Instruction Cache  8K Byte, 4-Way Set Associative (vs. 4K for 040)
Data Cache         8K Byte, 4-Way Set Associative (vs. 4K for 040)
Burst Fill Caches  16 Bytes R/W
On Chip MMU        Yes
On Chip FPU H/W    Yes

Since someone is bound to ask the obvious question, no they did not specify
which clock speed the MIPS and MFLOPS correspond to.  I know which clock
speed I would use if it were my table, though :-)

noah

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[e-mail]                                 Macintosh Hardware Design
...!{sun,decwrl}!apple!noah   (not the opinions of) Apple Computer, Inc.

------------------------------

Date: 20 Jan 1993 01:23:25 -0800
From: [e-mail] (Mark Rogowsky)
Subject: Power PC/68060/Taligent/Windows NT

Ah, I've been away sooooo long....

The 68060 vs. PowerPC and such discussion is one of great interest
to me as a power hungry user who doesn't want to buy a new CPU each
of the next 2-3 years. If you're like me, or, simply want a summary
of what's what, read on. The following contains no mention of MIPS
or SPECmarks (I don't even know what a SPECmark is), but contains a
fairly solid compilation of what is to come based on the words of
MacWorld, MacWeek, MacUser, PCWeek and THOSE WHO KNOW (i.e. my
"sources")

1) A 68060 Mac is likely for Q1 1994, as is the first generation
PowerPC-based Mac (using the so-called 601 version of the chip, a
single-chip implementation of the RS/6000 with Motorola's
88000-series mask (OOOH! Tech-talk!). The 68060 will be as fast, in
raw performance, as the 601. It will be faster in real use because
only "native-mode" RISC applications will fully exploit the 601.

An app like PageMaker, which uses Toolbox calls 90% of the time,
will run realy nice on the 601 because Apple is re-doing the Toolbox
for native mode. The System 7 rewrite, based on the "microkernel"
will most likely be the operating system for the PowerPC machines.
Taligent's OS is not going to be ready by Jan. '94. And, besides,
Apple has no plans to scrap System 7.x anytime soon. Like Windows
NT, Taligent will be the high-end solution "for those who need it"
for the forseeable future.

2) The PowerPC-based Macs will be much cheaper on a
price/performance basis than today's Macs. Finally, price parity
with Intel-based machines is in sight. But, there's a caveat. Until
YOUR application is rewritten for PowerPC native mode, it will run
at least partially emulated. That will slow it down a lot. The
rumors of $1200-1800 for a 601-based PowerPC Mac are, apparently,
true. This shouldn't be a stunner since Dell or Gateway will sell
you a 486DX2/66 for that price today.

PowerPC will be much faster that 486/66, of course, but by then
Pentium will keep Intel in the running. Pentium is the "586."

3) PowerPC gives Apple/IBM a chance to run past Intel in performance
because of RISC, etc. (whatever the etc. is). Basically, Pentium is
brilliant for Intel because it's software compatible with 486 while
significantly faster [although, to truly exploit Pentium at least
some recompilation is required. This is not as difficult as
rewriting for native PowerPC but is no walk in the park].

Power PC's product schedule is on track right now (amazing what
Motorola can do with a kick in the pants, where was this motivation
when we waiting eons for the 68040). It looks as follows:

601 - Available in limited quantities now. IBM has a PowerOpen (UNIX
alliance between IBM's AIX and Apple's A/UX) box on the schedule for
the first half of this year. Raw performance is rated at 10-20 that
of LCII. A meaningless stat, to be sure, but far faster than a
Quadra 950. Again, that's only with native mode apps. First 601 Mac
due in Jan 1994, probably on the 10th anniversary of the Mac, maybe
they'll even do a Super Bowl commercial (fingers crossed - Me).

603, 604 - The successors to 601, due one year later. These would've
been the first generation but for the pressures of getting it done
yesterday. 603 is a low-power variant designed for laptops and real
cheap desktops (sub-$1,000?). 604 should roughly double the
performance of 601 at equivalent speeds. They are looking at
eventually doing a 100MHz version. At that point, CPU performance
will cease to be a relevant factor in slowing down operations.
Video, SCSI, etc. will have to be greatly enhanced to keep up.

620 - The only 64-bit implementation on the board right now. Due
some time after 603, 604. Not a successor at all, but to be used in
workstation/server machines that do, say, 3-D rendering (are you
listening Silicon Graphics) or midrange-like server performance (a
multi-processor 620 may well be the eventual replacement for IBM's
venerable AS/400).

4) The CISC is dead, long live the RISC. Apple's product plans
change faster than ever these days but the last I heard is that even
the 68060 machines should be out of manufacture by late 1995 or so.
Apple is using the 060 as a bridge, much like the Quadra 900 was a
bridge until they could get 33MHz 040's in quantity. CISC based Macs
will not be manufactured when Bill Clinton is reinaugurated in 1997
(bet on it, just don't bet your life savings).

CISC has proven to be less limiting than people thought. When you
see appropriate software running on a 100MHz Pentium a year or so
>From now, you'll see why. But PowerPC is going to push the
performance envelope so much faster and, for whatever reason, at a
lower cost.

Apple believes it can sell RISC machines at prices lower than those
of today for an equivalent number of slots and amount of video
support, etc. The RISC machines will simply be eminently more
powerful.

5) Ch-ch-ch-changes take t-t-t-time. Let's figures on June 1995 as
the approximate date that RISC will be in absolute command. the
603/604 machines running microkernel System 7 with a native RISC
toolbox and native RISC apps (if you thought System 7 compatibility
was slow in coming, just wait. It will be a year minimum from the
601 intro to the point at which even all the major apps are running
native). That's over 2 years from now and cannot be compressed much.
The 601 machines present the buy-now-or-wait dilemma with a twist.
If you wait the year you'll have 603/604 for about the same dollars
(less perhaps) and now your apps will be ready to scream.

6) Taligent and Windows NT are guaranteed nothing. Microsoft should
ship NT soon but is working on a Windows variant that is a true
32-bit operating system (not a DOS shell) that doesn't require NT.
They don't believe everyone will shell out $400-500 for NT's
features when not everyone needs them. They do believe they can get
huge adoption rates for a Windows 4.0 that includes an OS (sounds
like the Mac, eh?)

Taligent, unlike NT, is a secret. Anyone can get the NT beta for
about $70 and know how it works, what it does, etc? That way, they
can WRITE SOFTWARE FOR IT!!!

Taligent hasn't released its specs, hasn't released a timetable,
etc. Microsoft advertises NT. Let's see, one open system (NT)
available to everyone, runs on existing hardware (well, sort of,
there are already millions of fast 486 and about 20 million more
will be purchased in 1993). One closed system (Taligent), secret,
unavailable except to the privleged few, no Microsoft monopoly and
marketing to promote it, runs of existing hardware (well, sort of,
maybe some 68040 Macs. Who knows for sure? They won't tell us. They
won't tell us what Taligent does while I can run NT today!).

This sounds a lot like MS-DOS/Windows/Intel vs. Mac. One dominates,
the other owns a niche that it holds with the iron fist of customer
loyalty. Apple and IBM each have 15 percent of the PC market right
now (approximately). That gives them less than one-third of the
potential NT vs. Taligent battleground as far as hardware-vendor-
sells-you-your-software logic.

Apple should scream "market share now," slash prices, outsource
manufacturing, and aim to double it's slice of the pie. Then we'd be
talking. The Mac's advantages are still manifold, even with Windows.
But, alas ... that's another thread ...

In the meantime, don't buy now, wait till Feb. 10 for sure. As for
pricing, just remember: A IIvx will have to cost a lot less than it
does now given that street price on the Centris 650 is projecting at
only a couple hundred more than current IIvx pricing. If the Quadra
800 isn't a great buy, you can always buy a Performa 600 (no cache,
no FPU, no matter), drop in a Daystar 33MHz 040 for $1350 street and
have a superfast Mac with the latest ROMs, a CD-ROM/DAT/SyQuest bay,
three NuBus slots and a nice power supply. You'd give up internal
video ($500-$1000 for a decent accelerated card) and fast SCSI
(which most people's hard disks don't have the throughput to use
anyway) and have, essentially, Quadra 800 performance for $4,000.
Apple knows this, too, and hopefully has worked out it's math
accordingly.

As always, comments/criticisms are welcome.

Cheers,
Mark Rogowsky
[e-mail]