Re: MacOSX on Prism...?
From: Benjamin Gawert (bgawert_at_gmx.de)
Date: Wed, 20 Oct 2004 20:32:17 +0200
Silas Denyer wrote:
>>> Not hugely original, but with the launch of the Prism and the
>>> *presumed* end of line drawing closer for MIPS/IRIX, what about a
>>> MacOSX port to SGI's Linux platform?
>> What for?
> To provide a (relatively) seamless transition from smaller to larger
> systems for graphics, animation, engineering, etc. application users.
> That used to be one of the key attractions of the SGI range (use
> desktop tools on an Indigo or Indy, run the big render or analysis
> parts of the job on the big iron).
But all these systems were binary compatible which is _not_ the case for
Itanium and PowerPC.
> To help position MacOS X as the "creative choice" in a "taller" range
> of markets,
Well, MacOS X still is the "creative choice" in a lot of markets. Especially
in the A/V market the Mac is a viable alternative...
> and bring SGI back into the creative limelight. To provide
> vertical market integration for SGI's advanced computing platforms.
How? SGI prooved multiple times that they can't handle the mass market. Apps
for MacOS/PPC won't run on MacOS/IA64 (if such beast would exist)...
>> Why should have Apple done that?
> Because the Indy was a direct competitor to the top end of the Mac
> product line at the time.
Definitely no. At least here in Germany the lowest entry Indy with nothing
included was around 7000EUR which was ~3x the price of a Powermac 8100 with
a good bunch of memory. For the price of the cheapest Indy You could buy a
nice Mac with an also very nice monitor and software. A comparable Indy was
more in the 12000EUR range which was way above what a comparable Mac went
I can't see how a unix workstation with 3x the price should have been a
direct competitor of mainstream computers like PC and Mac...
> Most users would choose a PC or an "other",
> and the Indy and O2 did for a while make relatively large inroads into
> the upper end of that "other" market.
Users choosed the Indy because it was the cheapest system offering 3D with
OpenGL. At that time neither Mac nor PC had comparable 3D solutions to
> At the time the Mac was the graphical platform of choice, but suffered
> from poor speed and a poor OS. Many users switched to Indys and later
> O2s - particularly with O2, SGI got an important march on Apple in the
> desktop Photoshop market with a proper, stable, fast system, precisely
> at a time when Apple was on the ropes.
Only a small number of customers bought Indies for Photoshop (which wasn't
really very stable btw). When the O2 came out 1996 it already was clear that
there won't be a next version of Photoshop any more.
> I've heard from a number of
> quarters that Apple exerted considerable pressure on Adobe to not
> release Photoshop 4 / Illustrator 6 on IRIX to try to get users to
> switch back to Mac as the "Photoshop platform of choice".
That's nonsense. Adobe cancelled their IRIX products simply because of the
very small customer base.
>> And how should that be possible with a port of MacOS X on Itanium?
>> Apple doesn't use Itanium but PowerPC which is not binary
>> compatible. That means applications for MacOS X on PowerPC won't run
>> on MacOS X on IA64...
> I agree with that. However I would make the observation that this
> would require software companies to recompile under MacOS X / Darwin
> for IA64 with appropriate compiler optimisations, not develop from
> scratch for the new platform. It wouldn't even be a "port" in the
> conventional sense of the word.
Having software to run efficiently under IA64 requires much more than just a
"recompile". VLIW needs optimizations to run fast, which would lead to heavy
>> Since Itanium is way different to PowerPC, most applications for
>> sure need more than just a recompilation but lot of work...
> Depends upon how the applications are written, to be quite frank.
For optimum performance rework is necessary in most cases...
> Anyhow, since most of the major apps have happily recompiled for MacOS
> X, and were previously recompiled from 68XXX MacOS to PowerPC MacOS...
No, no and no. First, the transition from 68k to PPC under conventional
MacOS was quite smooth for the customer. From the software side it was more
complicated. Most apps also needed some rework to run optimal under PPC.
Some apps also came as so called "fat binary" which contained an 68k version
and a PPC version. But most 68k programs ran still faster under the 68k
emulation on a PPC Mac than on a "real" 68k Mac.
As to MacOS X, it made a lot of problems for ISVs to port their apps from
MacOS 9.x and before to MacOS X. Due to the fact that MacOS X was totally
different from the old MacOS all apps needed a lot of rework. Nothing that
could be done with a simple recompile.
>> The Mac is much cheaper than the Prism, and in a different market.
>> Prism isn't really meant as a Photoshop machine...
> That's my whole point. That's like saying the Indy was in a different
> market from, say, an Onyx because it wouldn't run the "big iron" apps
> in any meaningful manner. The vertical integration is the important
> thing, surely?
Only if the market is there. Sadly, SGIs main market is _not_ the artists
market but the technical and governmental market, where Apple has almost
>> What for? Is it so difficult to realize that there's a reason SGI
>> goes away from MIPS/IRIX? Why the hell should they use money that
>> they gain from Itanium and invest it in a platform that has almost
>> zero applications and only a handful customers? MIPS is dying
>> because SGI can't make money out of it any more...
> OK, this is one of those points that gets argues a lot. I should have
> caveated my comments by saying "if they wanted to, and if lack of
> funds has been a handicap". There is considerable agreement AFAIK that
> IRIX is a tremendously stable, reliable, scalable OS, in contrast to a
> lot of other Unix flavours.
IRIX is by far not the most stable, reliable, and scalable OS. It's quite
similar in these aspects to the other major commercial UNIXes (AIX, Solaris,
HP-UX), better in some things, even worse in others. IRIX has some nice
features that were lacking in other operating systems, but these times are
over for lots of years now.
> Furthermore the MIPS architecture offers a
> whole host of advantages, if the development money can be put into it.
And that advantages would be what?
> There is still a large installed base, albeit much smaller than
> before. SGI can use the profits any way they like, I merely posited
> that they might wish to put them into their own world-class platform.
But if You would like to see SGI around in say 5 years from now such advice
would be highly (sorry) stupid. Putting money in a product that brings no
revenue is economical suicide. Most people forget that MIPS/IRIX isn't
really special any more. At the times it succeeded there weren't much
alternatives available. But these times are over. Gone. Definitely. Keeping
MIPS/IRIX alive by using money SGI earns with Altix/Prism is silly. If the
product can't bring enough renevue to at least a break-even point (no
loss/no profit) it's harming the company and just wasting money. Especially
if there is absolutely no sign that the product will bring any real profit
in the foreseeable future.
MIPS/IRIX is a thing of the past. It of course is part of SGIs history, and
should not been forgotten, but it's not competitive for the future any more.
The best SGI can do is following the Altix/Prism path which looks very
>> ...and far, far away from reality ;-) It would make more sense to use
>> Windows as Windowsxp has already been ported to Itanium, and it can
>> run x86 programs, too.
>> But both (MacOS and Windows) wouldn't make sense. SGI can tweak
>> Linux as they like, and they are free to put the IndigoMagic desktop
>> to Linux, too. There's really no need to pay a lot of money to buy
>> another operating system...
> No, there might be more point in Apple buying SGI though...
Well, no. As I said before Apple and SGIs market targets aren't really the
same. But Apple has very good chances to get into the technical market with
their G5 and Xserve systems. The problem seems to be more that Apple isn't
really interested in that, at least not at the moment.
> Darwin is Open Source, and has been (partly) ported to Itanium now, so
> SGI could still get involved with modifications, porting, whatever,
> and directly apply their hard-won IRIX expertise to improving the BSD
> codebase behind Darwin. That is a fundamentally different paradigm to
> going down the Windows route, which is resolutely closed-source.
Well, Darvin is just a part of MacOS X and lacks most things that makes
MacOS X that special. MacOS X is _not_ open source, it's based on open
source which is something different. Darvin on the other side is just a
modified OpenBSD, so instead of using Darvin SGI could simply use OpenBSD
which is available for IA64 already. But then, they already started with
Linux, which SGI also can tweak and improve as they want. So what?
Using an commercial operating system (be it MacOS X or Windows) only makes
sense if it also brings access to a) the market of that platform and b) the
applications for that platform. Since SGI doesn't use PowerPC or x86 they
better stay with Linux...
> The whole thrust of my (however off-the-wall) suggestion was really to
> prompt the debate about the vertical integration of creative markets.
Well, the idea was nice but is way off-realistic. And SGIs main market isn't
the creative market...
> SGI clearly don't have the money any more to compete with Macs on
> creatives' desktops (otherwise son-of-O2 would be serious contender),
> and the lack of software support (see Adobe etc.) killed the idea
> anyhow. But an alliance with Apple to create the "top part" of a
> complete creative solutions range could (as opposed to would)
> coneivably reposition SGI right back in the heart of a much wider
> creative marketplace.
Apple for sure has no interest in such alliance. They want to stay in the
market they currently are. Apple makes no efforts to expand to other makets,
and they barely can keep up with delivering for the demand in their current
markets (look at the waiting times for PM G5 and iMac G5). The Powermac G5
would make a very nice Workstation for CAD/CAM for example. It has a fast
CPU, very fast busses, and AGP Pro 8x for gfx. Sadly, there are no
professional gfx cards for the Mac. And the important CAD packages aren't
available for the Mac. Apple currently has no interest in this market and
does nothing to change that. Maybe one day they will, but not today.
SGI has nothing to offer that would be interesting for Apple. So it's highly
unlikely that they would be interested in such alliance.
And: the first Steve Jobs did when he returned was to get rid of the clone
manufacturer. They hardly would now license MacOS to anyone...
> I hope I've explained that a little better this time - the intention
> was not to propose complete lunacy so much as to consider a more
> unusual suggestion. Buying NeXT to underpin MacOS X wasn't necessarily
> seen as a particularly logical step either!
IMHO it was. The old MacOS was a dead end, and in it's last years was
technologically way behind everything else. NeXTs problem first was the very
slow hardware, and the prohibitive pricing. But NS/OS had lots of
interesting properties, and they could have done nothing better than giving
it a second live in MacOS X...