Could this be the future? Oh, I wish...
Consider the following: -
Steve Jobs makes a throwaway comment, which can be paraphrased as "You want performance - wait and see what we're going to release, you'll be blown away"
SGI is alledgedly considering a move away from its' own MIPS processors to PowerPC G5.
SGI has a market cap of under $550M, and cash/short-term investments of around $200M.
Apple has a market cap in excess of $7B, and cash/short-term investments of around $4B.
Apple appears to be indulging in a fit of whatever is the corporate equivalent of retail therapy, largely based around graphics and film/video post-production. Several of the products acquired are alledgedly 64-bit capable.
With these facts, what if:
Apple purchased SGI outright for $700M in a half cash/half paper deal: let's face it that $2B isn't earning anything sitting in the bank - well not as much as it could, and what's the point of having a half-decent market cap if you can't use it to acquire people who are down on their luck.
How does Apple benefit:
In one fell swoop, Apple acquires a mature professional market segment - although they are running someone else's technology.
Apple sets SGI's hardware engineering guys to work converting SGI to PowerPC, but combines this with Apple-centric "sanity checking" and skills-transfer. Components and sub-systems which yield best performance benefits make the cut and are adopted throughout Apple's pro range.
Apple cherry-picks SGI's software team and sets about merging the best bits of IRIX (64-bit capability or NUMA, perhaps) with the ease-of-use of OS X. Other teams are set up to integrate SGI's video-on-demand offerings on a QuickTime core.
Apple ultimately delivers four hardware product segments:-
Consumer & Education: eMac, iMac, iBook, iPod (and other i-devices)
General Corporate: Power Macintosh, Power Book, xServe
Specialist Corporate: revised SGI Workstation line, running G5 and OS XI (which is 64-bit pure)
Server: revised SGI Origin line, running G5 and OS XI, but with optimised graphics sub-systems modularised and made optional. Also the SAN and other storage products
Apple would then have an integrated product range, from SoHo and K-12 through small business through to corporate and specialist vertical visualisation and scientific.
Modularising Origin would allow Apple to deliver a range of general purpose "large iron" servers for data warehousing and the like. NUMA is really good for this kind of activity and both Oracle and Sybase have recently made very positive statments about their commitment to Apple, which can only help WebObjects in the application server marketplace.
Adding in the optional sub-systems to a revised Origin server will then give the type of renderfarm and visualisation benefits that so many contributors to these forum seem to feel they need.
The fact that Apple is also able to deliver a superb streaming media solution (based on QuickTime) is yet another benefit: MPEG-4 + 3G telephony + enormous I/O = a great package.
Apple would then operate the following cycle: -
Apple PMac goes G5 at MWSF 03
Apple TiBook goes G5 and SGI products relaunched at MWSF 04: OS XI released as 64-bit product
SGI products go to G6 at MWSF 05, iMac goes G5
PMac goes to G6 at MWSF 06.
Repeat ad infinitum until HP & Dell disappear into the mists of memory (sorry, just my poetic dream).
How does SGI (and their shareholders) benefit:
Let's face it: they are - as we English say - batting on a sticky wicket (it's a cricket term). Their marketplaces are being eroded by many other, including Apple, and they are know they can no longer afford the luxury of building their own processors (MIPS).
A galvanized HP/Compaq is hardly going to make things any better, and they are also unlikely to be a potential purchaser for SGI (A range that currently included IA-64, IA32 and Alpha processors doesn't need any more issues to deal with).
Sun are more than likely to just let SGI die a slow, painful death and then pick off the hardware/software guys they need.
IBM are a potential buyer, but SGI's people would need to adapt to a different culture than the one to which they are accustomed.
Put yourself in SGI's shoes
"Apple are just down the road - they believe in genuine innovation (not the plastic "innovation" talked about by our friends in Redmond) - they need SGI's engineering skills - our customers need continuity - and most of our people get to keep their jobs"
A marriage made in heaven, or just a contributor with too much time on his hands!
Discuss.
Steve Jobs makes a throwaway comment, which can be paraphrased as "You want performance - wait and see what we're going to release, you'll be blown away"
SGI is alledgedly considering a move away from its' own MIPS processors to PowerPC G5.
SGI has a market cap of under $550M, and cash/short-term investments of around $200M.
Apple has a market cap in excess of $7B, and cash/short-term investments of around $4B.
Apple appears to be indulging in a fit of whatever is the corporate equivalent of retail therapy, largely based around graphics and film/video post-production. Several of the products acquired are alledgedly 64-bit capable.
With these facts, what if:
Apple purchased SGI outright for $700M in a half cash/half paper deal: let's face it that $2B isn't earning anything sitting in the bank - well not as much as it could, and what's the point of having a half-decent market cap if you can't use it to acquire people who are down on their luck.
How does Apple benefit:
In one fell swoop, Apple acquires a mature professional market segment - although they are running someone else's technology.
Apple sets SGI's hardware engineering guys to work converting SGI to PowerPC, but combines this with Apple-centric "sanity checking" and skills-transfer. Components and sub-systems which yield best performance benefits make the cut and are adopted throughout Apple's pro range.
Apple cherry-picks SGI's software team and sets about merging the best bits of IRIX (64-bit capability or NUMA, perhaps) with the ease-of-use of OS X. Other teams are set up to integrate SGI's video-on-demand offerings on a QuickTime core.
Apple ultimately delivers four hardware product segments:-
Consumer & Education: eMac, iMac, iBook, iPod (and other i-devices)
General Corporate: Power Macintosh, Power Book, xServe
Specialist Corporate: revised SGI Workstation line, running G5 and OS XI (which is 64-bit pure)
Server: revised SGI Origin line, running G5 and OS XI, but with optimised graphics sub-systems modularised and made optional. Also the SAN and other storage products
Apple would then have an integrated product range, from SoHo and K-12 through small business through to corporate and specialist vertical visualisation and scientific.
Modularising Origin would allow Apple to deliver a range of general purpose "large iron" servers for data warehousing and the like. NUMA is really good for this kind of activity and both Oracle and Sybase have recently made very positive statments about their commitment to Apple, which can only help WebObjects in the application server marketplace.
Adding in the optional sub-systems to a revised Origin server will then give the type of renderfarm and visualisation benefits that so many contributors to these forum seem to feel they need.
The fact that Apple is also able to deliver a superb streaming media solution (based on QuickTime) is yet another benefit: MPEG-4 + 3G telephony + enormous I/O = a great package.
Apple would then operate the following cycle: -
Apple PMac goes G5 at MWSF 03
Apple TiBook goes G5 and SGI products relaunched at MWSF 04: OS XI released as 64-bit product
SGI products go to G6 at MWSF 05, iMac goes G5
PMac goes to G6 at MWSF 06.
Repeat ad infinitum until HP & Dell disappear into the mists of memory (sorry, just my poetic dream).
How does SGI (and their shareholders) benefit:
Let's face it: they are - as we English say - batting on a sticky wicket (it's a cricket term). Their marketplaces are being eroded by many other, including Apple, and they are know they can no longer afford the luxury of building their own processors (MIPS).
A galvanized HP/Compaq is hardly going to make things any better, and they are also unlikely to be a potential purchaser for SGI (A range that currently included IA-64, IA32 and Alpha processors doesn't need any more issues to deal with).
Sun are more than likely to just let SGI die a slow, painful death and then pick off the hardware/software guys they need.
IBM are a potential buyer, but SGI's people would need to adapt to a different culture than the one to which they are accustomed.
Put yourself in SGI's shoes
"Apple are just down the road - they believe in genuine innovation (not the plastic "innovation" talked about by our friends in Redmond) - they need SGI's engineering skills - our customers need continuity - and most of our people get to keep their jobs"
A marriage made in heaven, or just a contributor with too much time on his hands!
Discuss.
Comments
<img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> <img src="graemlins/oyvey.gif" border="0" alt="[No]" />
<strong>You call yourself a 'card carrying fanatic realist'?
<img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> <img src="graemlins/oyvey.gif" border="0" alt="[No]" /> </strong><hr></blockquote>
Sorry, England's performance in the World Cup is causing me to indulge in flights of fancy - you should have heard the stuff I was thinking of on Friday.
Penhead's "sanity check":
- Is SGI making any money?
I seriously doubt it.
- Does Apple stand to gain from SGI using PPC?
No they won't gain a dime. Remember, Apple doesnt make PPC, moto and IBM does. The ony way Apple would gain from SGI using PPC processors is if they made them themselves or somehow contributed to increased competition in the AIM space.
- Would the investments needed to incorporate SGI tech to PPC be worth it?
No it wouldn't. Apple is already working on a 64-bit OS X (it would be suicide not to).
With all the money Apple is throwing around these days on pro video markets, it stands to reason that they are more interested in working with mac in competition with SGI and others. What folly then to suddenly do an about-turn and buy them.
/penhead
You walk into Afghanistan and turn off the containment for a second, then turn it back on. The container will survive because it's made of bound neutronium. But everything within a hundred miles will be pulled toward the hole.
It will pucker the Earth. Like a hickey from God.
<strong>Now why would Apple want to do that?
Penhead's "sanity check":
- Is SGI making any money?
I seriously doubt it.
- Does Apple stand to gain from SGI using PPC?
No they won't gain a dime. Remember, Apple doesnt make PPC, moto and IBM does. The ony way Apple would gain from SGI using PPC processors is if they made them themselves or somehow contributed to increased competition in the AIM space.
- Would the investments needed to incorporate SGI tech to PPC be worth it?
No it wouldn't. Apple is already working on a 64-bit OS X (it would be suicide not to).
With all the money Apple is throwing around these days on pro video markets, it stands to reason that they are more interested in working with mac in competition with SGI and others. What folly then to suddenly do an about-turn and buy them.
/penhead</strong><hr></blockquote>
In response, SGI is a loss-making company sure enough!
Sure, Apple doesn't make any more money from SGI using PPC - although it does give IBM/Moto another reason to step up their game.
As for the investment to convert SGI tech to PPC: I would think the more appropriate question is: Is there an ROI that can be gained from making some of SGI's high-end technlogy work as part of a repositioned pro system line up. cc:NUMA (Coherent Cache: Non-Uniform Memory Access) is a big thing: it scales far better than SMP (i.e. over 32 processors), you want a render farm, how does 64 processors each working on their own bit of the job grab you. And from some of the text-based dribbling you get in some threads here, I would say that many of our fellow contributors would love to get their hands on the kind of graphics speed & capability that SGI can provide.
Also, I would damn well hope that Apple is working to deliver OS X as a 64-bit OS: I have been proposing in several of my contributions that Apple would want to deliver that sometime in 2004, so that revised versions of FCP, DVD Studio, Cinema Tools, etc. can all be offered as leading-edge tools at NAB 04. I don't think 2003 is viable as a target, but I do think the release of revised G5 Powerbooks will be remembered as the point at which Apple moved its' entire pro line to 64-bithood.
Would Apple be better for a more complete model range, including high-end servers and "real" workstations? I would think Yes
Are there a better bunch of guys for wringing every ounce of specialized hardware performance from a platform than SGI? Doubtful.
Would it be beneficial for Apple to acquire a global network of solutions-based pre- and post-sales personnel, especially if wants to widen the appeal of the brand? Yes.
Would it be beneficial to Apple to develop "proper computer" status in the eyes of the 95% of the universe do not share our "correct" view of the world? Yes.
$700M seems a small price to pay, and in a couple of quarters, $500M is going to seem almost irresistable to gain a business which has sales of $1B.
Let me reiterate and expand some of my points: -
If Apple/SGI became a reality:
Apple could develop a standard corporate computing product, with 2PCI slots and an 8XAGP, with a single G5 processor and one expansion bay, which it could sell very reasonably and fits the needs of most corporate business.
In parallel, it could develop a PowerMac Pro line with 5/6 slots, SGI graphics technology, dual and quad G5s and sell them as Apple/SGI workstations which would get the creative types (Photoshoppers, 3D mavens, video & film post-prods) vote.
And alongside all of this, it develops both a general-purpose server line for high-volume databases and application serving which can be transformed into highly functional renderfarms and streaming media servers, using NUMA and the kind of I/O for which SGI have market experience and Apple aspires and would have to spend 9 figures to develop.
This isn't so much about the size of SGI's market. It's about the depth of SGI's intellectual property, and the foundation that could provide for Apple to progress to a new level of performance. Sure, Apple could invent it all themselves, but didn't it get into trouble with that attitude before.
I don't know about you, but I got a tremendous buzz from seeing Final Cut Pro credited at the end of Episode II, but wouldn't it be great to see Apple technology credited as the post-production and 3D powerhouse of an entire epic rather than just a tool for pre-production in that market.
Rank speculation of course, and the rationale for buying an ailing competitor that might just pass on of its own accord is dicey. But Apple may want to prevent other competitors (MS, HP, SUN, IBM) from scooping up all of SGIs IP at a bargain price.
Blark <img src="graemlins/hmmm.gif" border="0" alt="[Hmmm]" />
[ 06-13-2002: Message edited by: Blark ] fixed typos
[ 06-13-2002: Message edited by: Blark ]</p>
As for Programmer's argument, there's not much for SGI to offer Apple. I'm sorry, but <img src="graemlins/surprised.gif" border="0" alt="[Surprised]" /> , Apple don't ship any 512-processor servers to my knowledge or any 4-ways either. Go and have a look <a href="http://www.sgi.com/servers/comparison.html"; target="_blank">at this page</a>, and tell me that Apple wouldn't benefit from converting this equipment to G5 (running at 1.6GHz+) and selling it as its' own.
Are Moto going to tell Apple how to do clustering (I'd rather they concentrated on getting some chips out personally)? Are IBM going to tell Apple how to do cc:NUMA (thus cutting their own throat)? and guess what? nVIDIA or ATI aren't going to tell Apple how to build SAN storage like <a href="http://www.sgi.com/products/storage/9400.html"; target="_blank">this</a> either.
Also, let's not forget who still owns the trademark on OpenGL.
It would take up to two years to fully convert/port/adapt SGI products and then release something based on all that tech hardware-wise.
If Apple were interested in the buy-the-superior-hardware-path they should have done it 2 years ago. Apple *has* done this with pretty much all pro software to date (FCP, DVDSP, Cinema Tools and whatever comes from Shake/Tremor and now RAYZ software).
I firmly believe Apple has taken over a substantial chunk of microprocessor development to suit its own needs and will use someone's fab to produce a future uber-PPC in 2003-ish. Maybe AMD or IBM (superior to Motorola who is contracting so much its amazing they do any business at all anymore).
This fits more with Apple's current stategy of PPC now PPC forever. SGI is largely dead in the water except in really big iron. Many many F/X houses are swithcing to PC hardware running Linux for a bulk of their 3D graphics work with render farms running various hardware (RenderBoxx, SGI).
SGI is losing its edge to Linux/PC hardware and hopefully some day to Mac OS X/Apple hardware. I see the ethereal "G5" as being the true SGI competitor for workstation performance hardware running under Mac OS X 10.2.x.
[ 06-13-2002: Message edited by: Nitride ]</p>
<strong>All this would be fine and dandy except for one niggling detail.</strong>
Only one! I must be closer to reality than I thought!
<strong>It would take up to two years to fully convert/port/adapt SGI products and then release something based on all that tech hardware-wise.</strong>
Don't forget one of the points I make: SGI is rumoured to be making the move to G5 anyway. All :-) that remains is engineering rationalisation and amending OS XI for the resulting product. If you put people under pressure and give them a challenge you could do that in 18 months, all you need is inspirational leadership (Steve Jobs) tight management (more Steve, with Avie et al) and great commited people (Anyone who works for either Apple or SGI has to be commited or potentially should be, there have to be easier walks to make).
<strong>If Apple were interested in the buy-the-superior-hardware-path they should have done it 2 years ago. Apple *has* done this with pretty much all pro software to date (FCP, DVDSP, Cinema Tools and whatever comes from Shake/Tremor and now RAYZ software).</strong>
Should have or could have? Big difference. Two years ago, SGI were $8 a share as opposed to $2.50 and Apple was still rebuilding credibility. Let's remember that 4-5 years ago, some of us were praying that Sony would buy Apple to keep it alive and inject some market savvy.
<strong>I firmly believe Apple has taken over a substantial chunk of microprocessor development to suit its own needs and will use someone's fab to produce a future uber-PPC in 2003-ish. Maybe AMD or IBM (superior to Motorola who is contracting so much its amazing they do any business at all anymore).</strong>
No contest!
<strong>This fits more with Apple's current strategy of PPC now PPC forever. SGI is largely dead in the water except in really big iron. Many many F/X houses are swithcing to PC hardware running Linux for a bulk of their 3D graphics work with render farms running various hardware (RenderBoxx, SGI).</strong>
My point exactly ("on a sticky wicket") - better they let Apple put them out of their misery, adopt/adapt their graphics technology to an OS XI framework, and allow Apple to fill out its' product range with their "big iron". Then F/X houses et al have a choice: a combination of systems built from various technologies/platforms or a holistic integrated solution thats been fully thought out from the ground up. You pays your money, you takes your choice.
<strong>SGI is losing its edge to Linux/PC hardware and hopefully some day to Mac OS X/Apple hardware. I see the ethereal "G5" as being the true SGI competitor for workstation performance hardware running under Mac OS X 10.2.x.</strong>
Not if my first point is true: what if Apple land up competing against SGI on a "G5" playing field.
<strong>[ 06-13-2002: All quotes from Nitride]</strong><hr></blockquote>
My point still holds true: buy them, preserve all that's worthwhile, fill out the product catalogue, use their pre- & post-sales support to support the consultant/dealer/VAR channel to sell a complete range of solutions, and (tx to blark), deny MS, Sun et al the opportunity to acquire Alias|Wavefront, get notional control of OpenGL.
Sounds like a plan to me :-) I'll be expecting a call from Cupertino any day now.
To quote Leonard Cohen: First, we take Manhattan, then we take Berlin.
[ 06-13-2002: Message edited by: Mark- Card Carrying FanaticRealist ]</p>
<strong>Isn't it simpler for Apple to buy US Congress rather than SGI? Then all the .edu market would belong to Apple until the next elections.</strong><hr></blockquote>
You're not from around here, are ya?
I'm not sure why people think SGI would be worth $700 million. Yes, SGI has a prestegious reputation and a few cool technologies. However, the high end workstation market is being, and eventually will be, taken over by the high end PC market. The rapid rate of PC/Mac performance increases make it increasingly difficult to justify extremely high end SGI based workstations. This is why NT and Linux based solutions have been making inroads as fast as they have. Hopefully Apple will start to enjoy part of this market based on the power of OS X, the new applications becoming available to the Mac because of OS X (like Maya, etc.) and because of the companies Apple is buying which target the Hollywood type of studios, etc.
Steve
<strong>I was wondering when this SGI rumor would surface again. It seems as though every couple of years this rumor pops up.
I'm not sure why people think SGI would be worth $700 million. Yes, SGI has a prestegious reputation and a few cool technologies. However, the high end workstation market is being, and eventually will be, taken over by the high end PC market. The rapid rate of PC/Mac performance increases make it increasingly difficult to justify extremely high end SGI based workstations. This is why NT and Linux based solutions have been making inroads as fast as they have. Hopefully Apple will start to enjoy part of this market based on the power of OS X, the new applications becoming available to the Mac because of OS X (like Maya, etc.) and because of the companies Apple is buying which target the Hollywood type of studios, etc.
Steve</strong><hr></blockquote>
I should point out that $700M seems extreme to me as well, but all I'm doing is taking the market cap as of today and assuming a bit (well OK, a lot) of inflation because of speculator activity once the company is deemed to be "in play". Personally, I would have thought $550M, 50% in Apple paper which has a known upside potential of 30% (based on yesterday's price and the 12 month high) would have sealed the deal. $700M was just a lazy way of saying how much it would cost to get rid of the problem, to everyone's satisfaction.
Again, I would say "don't focus on the workstations!": I would cannabalise the proprietary stuff that makes SGI stuff so good and adapt it for PMac, but would ditch the SGI workstation lines shortly thereafter.
My focus is the "big iron" and its' accompanying peripherals and software/hardware technologies. Again, please look at cc:NUMA - go to <a href="http://www.dolphinics.com/"; target="_blank">Dolphin Interconnect</a> and have a look at a whitepaper or two to see why I'm so keen (keep coffee by your side at all times though).
Dolphin's business is (I believe) based on IP it acquired from the estate of a US firm called Kendall Square Research (I swear the KSR-1 was the most freaky Star Trek machine I ever saw) which was founded by an ex-Data General employee (ah, Data General, my first machine, my first love, God I need help). BTW, I'm not alone, M$ is supposedly integrating NUMA into the Datacenter versions of XP server or whatever comes after W2K in that space.
This is high-tag high performance computing, which in a world where the questions are getting more complicated (protein folding, gene sequencing, individual phone numbers for everyone in the first world), is a pretty essential tool to have.
The billing and subscriber databases for a company like Vodafone (the partial owner of Verizon and the owner of probably the largest cellular network on earth) must be huge. And I don't even want to think about the computing power that will be required for the forthcoming (but much needed) computerization update for the UK's NHS. The only technology that will be capable of managing these huge MIS systems in a viable way is NUMA. If you want to have a longer conversation on this, e-mail me.
On the last point, I would have thought nothing makes it easier to be accepted into a market than just buying an entity which already has a brand image in that market and then relaunching it as your own based on a market dynamic that you create - btw, this is exactly what Apple is doing with Shake/Tremor & Silicon Grail - so why not but SGI, repackage, rebrand, reprice (in a realistic manner) and relaunch.
BTW, if you're Apple you have another reason to do all of this: consumer appeal. When Apple bought the Macromedia technology that became FCP, it didn't only amortize the purchase and subsequent development by selling FCP at $995, it abridged and abstracted the technology into a product which would create a USP for its' consumer offerings.
Apple could repeat this by introducing Technology X into the PowerMac Pro line (i.e. the SGI influenced PMacs) and then diffuse it down to the standard PMac line and then down to the i & e series systems over a period of time. The fashion business does this (e.g.Giorgio Armani/Emporio Armani), the automobile industry does it (Ferrari/Maserati/Alfa Romeo), and now the computer business does it.
The nett result of this is that the cost of development can be amortized over a longer period, which means that you don't need to recover the investment as quickly, which means you can lower the cost of acquiring that technology to its' intended consumer. In an ideal world, that would mean increased consumer interest, more market share and hopefully more sales and more profit.
But then again, an ideal world? Who am I kidding?
[ 06-13-2002: Message edited by: Mark- Card Carrying FanaticRealist ]
[ 06-13-2002: Message edited by: Mark- Card Carrying FanaticRealist ]</p>
<strong>As for Programmer's argument, there's not much for SGI to offer Apple. I'm sorry, but <img src="graemlins/surprised.gif" border="0" alt="[Surprised]" /> , Apple don't ship any 512-processor servers to my knowledge or any 4-ways either. Go and have a look <a href="http://www.sgi.com/servers/comparison.html"; target="_blank">at this page</a>, and tell me that Apple wouldn't benefit from converting this equipment to G5 (running at 1.6GHz+) and selling it as its' own.
Also, let's not forget who still owns the trademark on OpenGL.</strong><hr></blockquote>
IBM has the expertise to build massively multiprocessor machines, and last time I looked they still held the performance crown (1000+ processors, teraflop performance levels). And why would doing NUMA slit IBM's own throat? The whole RapidIO w/ on-chip memory controller idea is a NUMA scheme.
Having SGI use G5s would certainly help motivate whomever is building the G5 to spend more on it, but having Apple buy SGI just means Apple is again the only customer.
As for the OpenGL IP issue -- there are numerous companies which hold IP that OpenGL uses, that is possible because it is OpenGL and is directed and governed by the ARB.
<strong>
IBM has the expertise to build massively multiprocessor machines, and last time I looked they still held the performance crown (1000+ processors, teraflop performance levels). And why would doing NUMA slit IBM's own throat? The whole RapidIO w/ on-chip memory controller idea is a NUMA scheme.
Having SGI use G5s would certainly help motivate whomever is building the G5 to spend more on it, but having Apple buy SGI just means Apple is again the only customer.
As for the OpenGL IP issue -- there are numerous companies which hold IP that OpenGL uses, that is possible because it is OpenGL and is directed and governed by the ARB.</strong><hr></blockquote>
So your answer to NUMA is to say that IBM has the capability to build NUMA systems. No kidding!
My answer to that is that Apple needs to have control of its own destiny because, as Altivec shows, sometimes people and entities fall out. You and I are arguing at cross-purposes though: If IBM have the performance crown at 1000 processors, good for them. But is it performance that is optimized for Apple's potential customers' needs and is it scalable so that Apple can sell a 4-way box that operates in the same manner as a 64- or 256- way box.
Not many entities need a 256-way box, but I'll bet you even money that its a bigger market than the one for 1000+ processors. I'll give you a triple forecast and say that the same is true for 128 processors against 256, and 64 against 128.
I'm not saying that NUMA slits IBM's throat: I'm saying that IBM are not going give Apple a detailed diagram of how to build a NUMA box that is as good as the one that they have. Is that irrational? And if so, can you tell me how, because I'm confused. They're partners in a microprocessor development project, rather than being a permanent joint venture for every development under the sun.
Yes, Apple would still be the sole G5 customer: but a bigger customer with a bigger marketplace to develop and a bigger carrot to dangle.
I understand the OpenGL issue, but I believe my statement regarded the OpenGL trademark not all of the IP related to it. Surely, it helps Apple to be able to influence the direction of OpenGL and to have a commercial interest in its ongoing development, largely because it would give them early warning of the way the open-source project was developing.
This is no more than the case for 1394, and no one (sane) would argue that Apple should surrender its' ownership of the Firewire trademark without a quid pro quo of some sort.