MCCFR, you have several strong arguments. Since I'm too lazy to go look up SEC filings on SGI and crunch my own numbers (I've not done any diligence at all, due or otherwise on SGI since they bought out Cray, and I had a college professor assure me that "this means supercomputers on your desktop in five years!"), so I'm definitely arguing from a weaker perspective. Especially since my position is that costs are too great for value received.
So instead of doing real work, I'll take the lazy way out and go after the assumptions- if I can topple enough of them, then the subsequent conclusions based on them will topple as well.</strong><hr></blockquote>
OK, I don't mind the debate.
[quote]<strong>Assumption 1) Apple is working on a HPC product, or a product with HPC characteristics.
HPC is a very broad field, and traditionally ranges from individual vector processors (older Crays) to massively parallel machines (Beowulf clusters, ASCI White), and encompasses many things inbetween (Sun E15k). Prices are accordingly high- often starting in the $100Ks and scaling up to the millions.
There's no doubt in anyone's mind that Apple is interested in building computers which have better performance than their current machines. But a little bit better and orders of magnitude better both qualify as better than current offerings.
With prices between $800 and $4k, Macs are dirt-cheap compared to most all of SGI's offerings. In order to maintain these low prices, Apple adopts far more pedestrian, commodity parts.
Rather like the old MG or Triumph approach to performance- recycle existing parts in an exciting shell. (OK- cheap shot, but my last British car was my Dad's '52 MGTD... which doesn't go from 0-60 mph as it tops out about 55...)
Back to the argument, any HPC-like machine or technology that Apple employs (like a NUMA architecture) has to be cheap. Therefore, this must exclude many HPC characteristics which are too expensive to implement.
So if Apple is looking at HPC, it's looking at something really small-scale or cheap HPC, so we have minimal overlap with what SGI does... or charges.</strong><hr></blockquote>
A significant element of this statement is logically constructed, however you are omitting information or excluding certain concepts that for your own purposes.
I would have imagined that - if Apple had an interest in SGI - it would be to exploit the foundations that SGI have built. So cc:NUMA would be one of those cornerstones.
The challenge to Apple is not to produce a machine that is cheap, it is to produce a solution which scales both technically and financially and can deliver ROI through competitive differential advantage. I would argue that one of the SGI Origins (300, though it might be 3000) is the prototype for such an approach.
I am proposing that whilst SGI has an innovative HPC line, they are, becuase of their marketing and distribution model, unable to exploit anything more than a nominal market share.
Apple has the brains and the money to adapt this headstart so that new, innovative server platforms can be developed in a variety of product classes.
Small Business - General Corporate
=========================
AAPL have already released xServe, which competes against Compaq's DL380G2 in terms of spec and price.
From a technology perspective, this unit is probably where it needs to be: The addition of a "G5", with true DDR (hopefully at 400MHz+ running a 32-bit OS XI would just about be Nirvana.
Medium Business - General Corporate
===========================
I would have assumed that this would be AAPL's next server, released just after MWSF 03.
More than a simple "bigger" server, AAPL must release this to service 250-500 regular concurrent users. My guess would be a system that can go from 2 to 4-way as standard, and with a true SCSI (six drives) array, internal expansion bay for DLT/AIT, plus six HyperTransport/PCI-X compatible expansion slots.
AAPL would be pitching this machine into the ML370G2 class, for file/print, but with the advantage of being scalable to 4-way and a 64-bit OS could easily pitch against IBM's x255/X360 (not 64-bit, but Xeon MP which given Itanium's performance is no bad thing).
Large Business - General Corporate
===========================
Joint-top Billing with G5 TiBook (MWSF 04).
Keyword: Fabric
Key Technology: InfiniBand
A modular server product comprising up to 4 core modules (each delivering either 4-way or 8-way processor modules) with the ability to deliver 64GB in RAM and six HyperTransport slots in each core module, additionally: -
Core storage module (4 x SCSI-320 drives)
Sharable (two host) expanded storage module (2 x 7-slot , deploying up to 14 x 240GB+ fibre channel native drives (i.e. 3.2TB formatted) or 7 x 240GB + 1 x tape module (DLT/AIT) or 1 x Tape Changer.
Sharable (two host) expansion module delivering 12 HyperTransport/PCI-X or slots or better.
Target Market : Database + Apps Server + Small Scientific/Visualisation
Current key competitor: IBM X440 (again, Xeon MP and 32-bit only)
All I can say so far is, where do sign !
But this is where my plan gets interesting.
Large Business - Scientific & Visualisation
===========================
Time: SIGGRAPH 04 or MWSF 05 or NAB 05
Target Market : Data Warehouse + Apps Server + Large Scientific/Visualisation
First thing up is Visualisation Module, called something squiffy like Infinite Reality, containing the latest evolution of all of SGI's visualisation technology. Also delivers optional digital video I/O interface, such as SDI and Gigawire. (SGI guys front and centre). Fits into Infiniband-fabric server announced at MWSF 04.
Announced to cheers from the multitude, SJ shows off the first AAPL implementation of SGI's Reality Center, with an immersive VR screen the width of the Moscone Center stage, SJ braves the attack of several dozen Empire fighters before destroying the Death Star.
But surely that wasn't all done with just 32 G5 processors
SJ now announces one last thing: A router box for the management of a single coherent shared memory which connects to an implementation a 12X 30Gbit (3.75GB) InfiniBand interface.
A few bits of software jiggery-pokery (SGI's software engineers front and centre) and up to 64 8-way modules (which take up no more than 3U of rack space each) can be connected together to deliver a 512-way MPP system that delivers 4TB of coherent shared memory (so normal programmers can use it easily).
Allowing for a 5U router module for each rack, plus UPS and other ancillaries, Apple would have a solution which could scale from a single 22U rack (servicing several hundred database users or a hundred visualisation users or a couple of thousand broadband streaming customers) to (dependent on storage and tape backup requirements) 10 or more 42U racks (servicing several thousand data warehouse users or several hundred visualisation users).
This is the core of my server argument: AAPL can evolve and scale a single highly marketable product (4-way, 64GB, 6 PCI-X) with an enormous audience into a highly glamorous, highly profitable, HIGH PROFILE product which will make Sun, IBM, and HP/Compaq look around desperately to create some kind of viable response.
In terms of density, protected investment, consistency of support , this would be a world-beating range.
[quote]<strong>Assumption 2) SGI is working on a PPC-based machine. Not provable one way or the other, and if not true, dampens a lot of the fire since this assumption is crucial to transition costs. If true though, this machine is more likely to be a million dollar server than a $1k blade. Again, we're not guaranteed that costs of this technology fall into line with what Apple can afford to charge, or even be the kind of technology that Apple's really interested in at the moment.</strong><hr></blockquote>
I have to say this argument ignores commercial reality: when VW purchased Bentley, did they say "We'd better be careful, because VW can only afford to charge $20K to sell to its traditional audience" ?. I think it is far more likely that they said:-
[quote]<strong>"Here is a company whose brand image and values will be commercially exploited by either DaimlerChrysler or BMW if they get hold of it. We can afford to make Vickers an offfer and enhance our image in the marketplace, which will hopefully lead to more sales for our other brands. Oh, and they sell for a fortune to a loyal audience with more money than sense, so we can make a profit!".</strong><hr></blockquote>
Replace DC and BMW with Sun or Compaq, replace Vickers with "SGI's shareholders" and I would have thought that was a position that Apple's board could take.
As for the G5 thing! Have I not been clear enough on this point - it's a rumour, that's why I occaisonally use the phrase "If The Rumour is true...".
It probably is not true, but then again it might be given the MIPS market positioning thing I mentioned a few posts back.
Maybe SGI have taken the view that MIPS are going in a direction (embedded, mobile, games console) that is incompatible with the mainstream markets that SGI have to pursue, they need to align themselves with another high-performance processor design which has a long-term future. I don't know, it could all be the product of an overactive imagination which was never apparent in my school years.
[quote]<strong>Assumption 3) SGI is sufficiently inexpensive that Apple can pick it up without too much sweat. MCCFR has done a very good analysis demonstrating that this is probably true. The only real hole I can knock in here is that even though it's within Apple's budget, it still represents a sizable expenditure which may be better spent elsewhere.</strong><hr></blockquote>
Might be!, Might not be! If my aunt had a moustache, she would have been my uncle, Hold on, my aunt does have a moustache, but I know she's not my uncle.
I think all of these questions can be solved with due dilligence, although - especially with my aunt - I'm glad I'm not responsible for any part of it.
[quote]<strong>Assumption 4) Synergy in Apple's and SGI's video editing and 3D graphics markets.
While true to a degree, key apps like Maya already exist for MOSX, so other than control over the direction of Maya, the major benefit of Maya-on-the-Mac already exists. More importantly, the key infrastructure of generating video editing software (iMovie, Final Cut Pro) is already in place and running on both today's and yesterday's Macs. Whether software like this would run so much faster on future SGI hardware or future Apple hardware as to justify the cost of making it happen is up in the air. I don't know if high-end markets like television stations are willing to fork over the $100k for a high-end Final Cut Pro on SGI-class hardware, or whether $4k Apple hardware is sufficient to get the desired work done. The one thing I can easily state is that the demand curve clearly shows that more $4k boxes will sell than $100k boxes.</strong><hr></blockquote>
The art of a good Scrabble player is someone who can take another player's word, add a few letters, taking in a double-word score as they go.
So, "The major benefit of Maya-on-the-Mac exists" becomes "The major benefit of Maya-on-the-Mac exists for as long as SGI continues to be independent of Microsoft or another predator".
Equally, "I don't know if high-end markets like television stations are willing to fork over the $100k for a high-end Final Cut Pro on SGI-class hardware", becomes "I don't know if high-end markets like television stations are willing to fork over the $100k for a high-end Final Cut Pro on SGI-class hardware, but they might handover $100K if they got online compositing, plus editing, plus whatever else".
As for "Whether $4k Apple hardware is sufficient to get the desired work done", I'm not sure I'm offering a $4K solution, but I think I might be offering a powerful $20K+ solution.
BTW, can someone verify this equation: (AAPL*QT6)+ SGI + AVID = market dominance of professional and corporate audio production + broadcast news/film/video/DVD post-production + class-leading broadband streaming of the end product.
Yours for a $750M combination of cash + shares, plus restructuring expenses and ongoing investment.
Commentary, in black and white, but may be interrupted with brief flashes of color:
Re: Apple and HPC
[quote]
A significant element of this statement is logically constructed, however you are omitting information or excluding certain concepts that for your own purposes.
<hr></blockquote>
Yup- an argument is an intellectual process, not the automatic gainsaying of anything the other person says. (Bonus points for whoever can name where that line came from!) Besides, isn't one of the goals of rhetoric to not destroy one's argument by contradicting it, something I admit I'm not always very good about?
[quote]
Small... Medium... Large Business
<hr></blockquote>
One of the problems with this is that it presupposes Apple's interest in "climbing up the food chain" to play in Medium and Large Business computing. Xserve is pretty limited in its target market (small server), and worse, the Medium and Large market is largely saturated as is with competitors like IBM, Dell, Hewlett Compaqard and Sun. Further, the technology Apple is using or could use as proposed (with SGI buyout) is not sufficiently compelling to grab a significant chunk of Medium and Large IT budgets.
I don't think that it's a stretch to say that Apple has stayed out of these markets since the return of Jobs because Apple doesn't want to play in these markets, instead concentrating heavily on the consumer and desktop publishing markets. Now, of course, they're trying to expand the "creative" market by pushing into video editing. As an aside, 3D modeling seems to be coming along for the ride, but doesn't appear to be a major focus of current Apple strategy.
If you look at the targeted markets mentioned above-
[quote]
Target Market : Database + Apps Server + Small Scientific/Visualisation
<hr></blockquote>
Most of these don't line up well with Apple's strengths. Database in particular- more people run Sybase on Suns for a reason. Oracle 9i is just now being ported to MOSX, and I'll go out on a limb here and say that the people currently buying Suns, Dells, IBMs, et. al. to run Oracle on will keep buying Suns, Dells, IBMs, et. al. because for at least the near term, Oracle has a proven track record on them. These vendors have demonstrated DB quality that Apple will never have until they've shipped DB quality machines long enough to demonstrate that quality.
Apps servers are all environment and application specific. Again, Apple is late to this party, and so lacks any first-mover advantage.
Scientific and Visualization here has at least some overlap with Apple's vision (BLAST for example), but one of the big trends here is for a distributed computing model- one where the XServe has as many advantages as the competition.
Expending effort to "crack" these markets will simply drain Apple's coffers and distract from the areas Apple wants to compete in- consumer, education, and creative markets.
[quote]This is the core of my server argument: AAPL can evolve and scale a single highly marketable product (4-way, 64GB, 6 PCI-X) with an enormous audience into a highly glamorous, highly profitable, HIGH PROFILE product which will make Sun, IBM, and HP/Compaq look around desperately to create some kind of viable response. <hr></blockquote>
I don't think this is very realistic. As I mentioned before, this is a very crowded market to debut a product like this into. One might even say saturated. Chances of success are pretty slim without some major "killer app" to drive it, be that price, performance, or software-based.
Re: SGI working on a PPC-based machine:
[quote]I have to say this argument ignores commercial reality: when VW purchased Bentley... <hr></blockquote>
I disagree. I think that's not a very good analogy for many reasons, not the least of which is the fact that VW is already composed of several different brands, something Apple is not. So adding a new brand is a different concept than buying out and shutting down a different company to leverage yourself into a new-to-you market.
Conversely, looking back at other arguments, your proposed strategy is more one of "SGI... by Apple" instead of "Apple... by Apple." So perhaps what you're really proposing is melding SGI and Apple together to create a different company than either is today. Whether this corresponds to the direction that Apple is currently headed in or not is an interesting question. MCCFR, if you choose to respond to this post, could you please expand a bit more in this line? Thanks in advance...
[quote] As for the G5 thing! <hr></blockquote>
Yeah- it's an Emeril Live! thing (Bam!), the argument is kicked up a notch if the rumor is true. If not, then SGI just looks more overpriced. Regardless, it's a key assumption in the cost equation.
Re: SGI is inexpensive...
[quote]If my aunt had a moustache <hr></blockquote>
Sorry to hear about your aunt. Personally, I've decided not to make fun of hairy people anymore because the older I get, the more (and in odder places!) hair I get. It's one of those "glass houses" things...
Re: Synergy...
[quote] The major benefit of Maya-on-the-Mac exists for as long as SGI continues to be independent of Microsoft or another predator <hr></blockquote>
In context, this actually isn't that big of a problem. One key assumption, that A|W is independent of M$ or another predator is self-evident by the fact that, despite the "bargain basement price", no one wants to buy out SGI and "Chainsaw Al" it into its components. In other words, SGI isn't in play, and it isn't because no one wants to spend $500-750M in cash or equivalents to do it. Which clearly shows that A|W isn't worth $500-750M plus the trouble of chopping it out of SGI. So that's a natural defense against it getting bought out.
Even more, small companies that are under control of "predators" still produce software for the Mac market... because there is money to be made doing so. Avid, for example, despite attempting to leave the Mac market, still makes hardware and software products for it. Avid very much shot itself in the foot for this, and is not in very good shape for it. Rather a lesson for anyone who would want to buy out A|W and shut down a very popular platform for its premiere (heh! pun intended) product...
<strong>This is the most wordy thread I have ever witnessed. I think i just wasted about an hour reading idle ramblings and such. Keep it up.</strong><hr></blockquote>
Well, at least that's one thing this thread has achieved! <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
As for HOS, will respond as soon as I've had a chance to consider your arguments, except on one point on which I agree.
What I object to about middle-age is the way the hair migrates from your head, to your ears and nose! If there is anything more annoying than one of those thick, short, stubble hairs rooted in your earlobe, I have yet to find it.
I admit this is not about Future Hardware, but I would buy any gadget that dealt with this issue - Remington clippers don't do the job, cos the hair is too short.
i think it was monday, but i only got 2 pieces of email. a monthly newsletter from sgi, and some dumb AD-mail from apple. now dont worry, i dont read into that crap, but after reading this thread all weekend it was good for a laugh. just thought id share.
[quote]<strong>Yup- an argument is an intellectual process, not the automatic gainsaying of anything the other person says. (Bonus points for whoever can name where that line came from!) Besides, isn't one of the goals of rhetoric to not destroy one's argument by contradicting it, something I admit I'm not always very good about?</strong><hr></blockquote>
On the bonus question, I believe its Monty Python (the argument sketch), but I cannot remember who the performers were. Palin & Cleese, perhaps or Palin & Jones. maybe someone could remind me.
I don't object to you making an argument, but I think that they should at least vaguely observe some logic.
An example of what I'm complaining about would be: Apple currently makes mainstream systems using commoditized components which are sold at mainstream prices, therefore Apple cannot exploit the HPC computing market because they would have to retain a mainstream business model and could not therefore build a high-end system.
This, to be blunt, is abject nonsense of the most absurd kind. Nothing is written in AAPL's articles of association that dictate that change either by way of market segmentation or product & market diversification is forbidden. So the only thing that prevents AAPL from improving its brand image in the corporate, scientific and specialist user markets is the will to do it, whether from scratch or by a process of acquisition.
One of the problems with this is that it presupposes Apple's interest in "climbing up the food chain" to play in Medium and Large Business computing. Xserve is pretty limited in its target market (small server), and worse, the Medium and Large market is largely saturated as is with competitors like IBM, Dell, Hewlett Compaqard and Sun. Further, the technology Apple is using or could use as proposed (with SGI buyout) is not sufficiently compelling to grab a significant chunk of Medium and Large IT budgets.</strong><hr></blockquote>
Using this argument, every marketplace that AAPL competes in is saturated with everything from Tier 1 suppliers like IBM, HP/CPQ and Dell through mail-order to a million Main St. PC shops with beige boxes to sell.
As for making a compelling product, I would dispute that aggressively and will do so throughout this post.
[quote]<strong>
I don't think that it's a stretch to say that Apple has stayed out of these markets since the return of Jobs because Apple doesn't want to play in these markets, instead concentrating heavily on the consumer and desktop publishing markets. Now, of course, they're trying to expand the "creative" market by pushing into video editing. As an aside, 3D modeling seems to be coming along for the ride, but doesn't appear to be a major focus of current Apple strategy.</strong><hr></blockquote>
HOS, a career in spin doctoring awaits you here in the UK should Tony Blair's Richlieuesque Communications Director choose to go and get a proper job.
From personal experience, I would say that the reason for AAPL staying out of those markets is not that it chooses to, but that its invitation to the right parties "got lost in the post" because it didn't take the "privilege" of working with major corporates seriously.
I have a shameful secret here: Having worked with a major corporate in the oil sector under contract delivering Mac projects and support for three years from 1992, I was ultimately involved in migrating several key elements of its business away from Mac to Win95/O95 (client) and NT 3.51/4.0 (server). Why? Because AAPL in the UK didn't take the threat of an aggressive approach from MS to the customer seriously, and AAPL at a global corporate level (the era of Sculley and Spindler) didn't take charge of the situation.
That's as much information as I can share on a public forum, but you can read what you want between the lines and its probably true.
Working with major corporates doesn't necessarily make you the profit you want, but it does keep your profile afloat for smaller corporates and SMEs who like to go with the flow.
Most of these don't line up well with Apple's strengths. Database in particular- more people run Sybase on Suns for a reason. Oracle 9i is just now being ported to MOSX, and I'll go out on a limb here and say that the people currently buying Suns, Dells, IBMs, et. al. to run Oracle on will keep buying Suns, Dells, IBMs, et. al. because for at least the near term, Oracle has a proven track record on them. These vendors have demonstrated DB quality that Apple will never have until they've shipped DB quality machines long enough to demonstrate that quality.</strong><hr></blockquote>
The reason that companies run Sybase or Oracle on Sun Starfires is a relatively simple combination of factors: it's the only agressively marketed 64-bit general purpose business applications server with a comprehensive ISV business applications portfolio that runs a robust "open" operating system.
If AAPL delivered a modular server platfom, scalable from 2-512 processors, and with any visualisation technology made modular and, by implication, optional, it would have a solution that was (at worst) highly competitive in the marketplace or (at best) a definitive technology platform which could outscale and outperform any platform in its class.
Performance, combined with strategic applications like 9i or Adaptive Server (which is also being ported to OS X) and supportive/assertive developer relations marketing will lead to ISV's who have not been traditionally associated with Apple - like SAP, Siebel or Documentum - porting their applications to OS X.
AAPL would also have the advantage of being able to deliver a viable range of general business desktop/laptop systems which contributes to a lower TCO argument against Sun (and other Unix platforms) and counters the same argument from the WinXXX market.
If you accept all of these aspirations, then AAPL would then be ideally placed to deliver consistent 64-bit performance in all major sectors (client, workgroup, server room and datacenter) and a consistent user experience across all of those segments. This last point is more important that might first appear, but I feel the justification belongs later in this post.
[quote]<strong>Apps servers are all environment and application specific. Again, Apple is late to this party, and so lacks any first-mover advantage.</strong><hr></blockquote>
I was not referring to narrow application servers, but to web application services where AAPL has WebObjects as a well-respected product. However that respect does not necessarily translate into sales: This, I suspect, is partially because AAPL is not seen as being aligned with the mainstream industry, which a broader product range and a more comprehensive ISV business applications portfolio would undoubtedly address.
In addition, AAPL could/should try to broaden the functionality and appeal of WebObjects, so that it is more capable of answering the needs of the wider application services market, thus providing a viable alternative to .NET which MS will probably use as a sledgehammer to crush any development language that is not developed at MS, such as Java or Objective-C. Partnership with MS here is not practical as it would lead to junior partner status for AAPL, but marketing WebObjects as an "open" alternative to .NET will give dissentors a mainstream place to go.
[quote]<strong>
Scientific and Visualization here has at least some overlap with Apple's vision (BLAST for example), but one of the big trends here is for a distributed computing model- one where the XServe has as many advantages as the competition.</strong><hr></blockquote>
This may be true in some areas, but it is not a universal truth. Speaking from my most recent flirtation with the Oil/Petrol industry, I noticed that there were more Unix servers (of all complexions) in the primary server suite than the visit before that. I can't help but feel that commercial science will always have a tendency to gravitate to centralised computing because of support arrangements.
[quote]<strong>Expending effort to "crack" these markets will simply drain Apple's coffers and distract from the areas Apple wants to compete in- consumer, education, and creative markets.</strong><hr></blockquote>
You're spinning again! Wants to compete in or has to compete in ? I feel you're turning a necessity into a virtue. Quite frankly, this is the equivalent of a car manufacturer like Rover saying it's not interested in the North American marketplace, whereas the reality of the situation is that it got its butt severely kicked around the last time it tried so it went off with its tail between its corporate legs.
To paraphrase Oscar Wilde, AAPL so far has done a fair job of convincing two audiences who "know the cost of everything" i.e. consumers and education. But in the former case, there is still a lot of work to do to increase user base (whether expressed as market share or absolute numbers) and the latter case is dependent on the political priorities of a given state or school district.
In addition, both of these markets are dependent on the health of the economy: the former is dependent on discretionary expenditure and the latter is dependent on a vibrant tax base.
Another issue is that AAPL's success in the K-12 market is not universal: In Europe, much of that business goes into the WinXXX market, because (rightly or wrongly) that is seen as an affordable option for students in higher education and seen as the de facto standard in business.
I concede that AAPL will always see the "creative markets" as its home turf. However, I am sceptical that we can always rely on that market as a cash cow.
The laughter that used to accompany discussions of Windows 3.x or Windows 95 being used for serious creative work has become a little more nervous in the face of WinXP (Fisher-Price toy interface or not) and the perceived performance gap that now exists.
The fact that the New Media industries have never bought the "Macintosh Uber Alles" party line to the same extent as print media is telling. The desire for colour management is not quite as extreme in the web world and New Media tools evolved in the WinXXX universe more quickly than on Mac resulting in an obvious bias towards WinXXX.
The market that remains is that for A/V post-production and workflow, which as you rightly point out is the market that AAPL are obviously lining up for dominance given the recent acquisitions. And I genuinely hope they succeed!
However, I believe that the time has come for AAPL to try and stretch beyond the status of a "creative ghetto" system or a system for domestic consumers with an eye for design flair.
I want to see AAPL leverage a scalable OS on a variety of system segments all the way from Home to Enterprise!
I want to see AAPL create a set of USPs for itself, which allow customers (both existing and potential) the opportunity to create USPs for their businesses!
Where there is a business or scientific challenge, I want to see AAPL being considered as a potential solution to that challenge!
I am passionate about this product and its future, and I believe that the future for AAPL is to get back into the corporate market.
[quote]<strong>I don't think this is very realistic. As I mentioned before, this is a very crowded market to debut a product like this into. One might even say saturated. Chances of success are pretty slim without some major "killer app" to drive it, be that price, performance, or software-based.</strong><hr></blockquote>
First off, I don't think we should describe price or performance as "killer apps", otherwise the whole benchmark or MHz thing becomes the killer argument whereas it should be an equal factor in the evaluation process.
Defining a killer app is difficult, because that app varies from industry to industry. But let's explore a few possibilities which I will ask as questions because I'd like to hear a few different opinions.
Solution A
======
Looking at server consolidation, which is a vogue at this time, could AAPL use the skills of a Connectix and/or VMWare to develop a partitioning solution for their servers so that a single 32-way server could be utilised to run Win2K on two, three or four partitions (let's say two x 1-way, two x 2-way). 64-bit OS XI could be run on 2 x 8-way partitions 1 x 4-way partition and 1 x 2-way, and Linux could run on the remaining 8-way partition.
Would Win2K (or it successors) running under emulation on a 64-bit platform be performance-competitive when compared with running native under Intel or AMD 32-bit architectures? Can someone answer that?
I'm not wholly keen on this as a "killer app", largely because it feels uncomfortable at a religious level giving MS another way to implement its "solutions". However, it would recognise the fact that many customers are committed to using using MS for Exchange and AAPL do not have an enterprise e-mail system that they can put up as a solution.
Solution B
======
Another popular corporate technology (certainly since Win2K) is Citrix MetaFrame, which - for those who are unaware - is a solution for delivering thin-client Windows sessions either over a) slow WAN links or b) on clients which no longer meet minimum equipment specs to run Win2K or c) for corporate apps which are considered to be appropriate for centralisation which is deemed to be a cost-saver.
In reality, MetaFrame takes a significant influence from windowing technologies such as X/Windows, where I think I'm correct in saying that presentation was the preserve of the "client" system but computation was the preserve of the "server".
The problem with MetaFrame is that it is quite stunningly expensive to implement (both in terms of license fees and the quality and depth of infrastructure that is required to run the solution).
So my question is: Could AAPL adapt an evolution of X/Windows (Unix core after all), so that personal productivity apps (such as Office) run as purely client apps, but business management apps (e.g. SAP) could run as windowed apps? And if this were technically feasible, could it be delivered with typical Apple flair (as opposed to MetaFrame, which is not as pretty as it could be) and at a cost which significantly undercut MetaFrame?
If you combined A and B with cost-effective SMP or HPC systems, AAPL would surely have a set of tools which constituted "bait" for corporate IT functions to wean themselves off Intel-based infrastructures.
If you accepted all of my arguments in this post, then really the only major tool that AAPL would be missing from its arsenal to establish a comprehensive range of IT solutions at the end of this process would be an Exchange-killer. I have to admit on this point, I have no opinion or way forward - my opinion of Notes is pretty low and my opinion of Outlook (on Windows) relatively high, so I would appreciate some input on how that issue could be resolved.
My own thought was whether Sun would ever consider making some of the iPlanet products available under OS X, but I have no knowledge of these at all (except evaluating the product that became the collaboration server when it was Collabra) so would appreciate some input.
[quote]<strong>I disagree. I think that's not a very good analogy for many reasons, not the least of which is the fact that VW is already composed of several different brands, something Apple is not. So adding a new brand is a different concept than buying out and shutting down a different company to leverage yourself into a new-to-you market.</strong><hr></blockquote>
Ok, but your original argument which caused me to use that counter was something like: Could Apple sell outside of its traditional marketplace?
My answer to that is still "Yes", and if you dispute my first analogy than I will simply have to use VW again and point you at that company's soon-to-be-released Phaeton. Whether a company (whose name translates as "people's car") selling a model for £60K+ is ironic, I will leave for others to discuss. But it is outside of VW's standard market.
[quote]<strong>
Conversely, looking back at other arguments, your proposed strategy is more one of "SGI... by Apple" instead of "Apple... by Apple." So perhaps what you're really proposing is melding SGI and Apple together to create a different company than either is today. Whether this corresponds to the direction that Apple is currently headed in or not is an interesting question. MCCFR, if you choose to respond to this post, could you please expand a bit more in this line? Thanks in advance...
</strong><hr></blockquote>
I'm not a marketing guru, so I'm a little unsure of the best way to proceed here, but I'll give it a bash.
I think the relationship that I would choose for SGI would be analogous to the relationship that AMG has with Mercedes-Benz or Alpina has with BMW. I believe that the core technologies should be marketed as an Apple solution, but wherever SGI technologies have been deployed to enhance the core solution, that the enhancement is recognised along the lines of "Apple, tuned/enhanced/whatever by Silicon Graphics" .
This would stress the foundation of the system, but also continue the exploitation of SGI brand and provide continuity for customers and employees alike.
Does this answer your question?
[quote]<strong>Yeah- it's an Emeril Live! thing (Bam!), the argument is kicked up a notch if the rumor is true. If not, then SGI just looks more overpriced. Regardless, it's a key assumption in the cost equation.</strong><hr></blockquote>
OK, let's accept that as a given.
[quote]<strong>In context, this actually isn't that big of a problem. One key assumption, that A|W is independent of M$ or another predator is self-evident by the fact that, despite the "bargain basement price", no one wants to buy out SGI and "Chainsaw Al" it into its components. In other words, SGI isn't in play, and it isn't because no one wants to spend $500-750M in cash or equivalents to do it. Which clearly shows that A|W isn't worth $500-750M plus the trouble of chopping it out of SGI. So that's a natural defense against it getting bought out.</strong><hr></blockquote>
And I accept that for the moment, but there may come a point where SGI is an animal so weakened by fighting its corner that the opportunity for predators becomes irresistable. My worry is what happens thereafter.
[quote]<strong>Even more, small companies that are under control of "predators" still produce software for the Mac market... because there is money to be made doing so. Avid, for example, despite attempting to leave the Mac market, still makes hardware and software products for it. Avid very much shot itself in the foot for this, and is not in very good shape for it. Rather a lesson for anyone who would want to buy out A|W and shut down a very popular platform for its premiere (heh! pun intended) product...</strong><hr></blockquote>
I accept your Avid argument totally, and accept that it might be a lesson for a potential A|W predator.
However, if a predator's aim was to slowly "disincentivise" the Macintosh userbase, a great way to do it would be to disconnect parallel development of the WinXXX and Mac versions of a product. Then, slowly accelerate the WinXXX development so that it was superior in terms of functionality and performance, and eventually the customer makes the decision for you.
I can't believe I read the whole thing (actually I didn't, I just scanned it). One question popped out at me though:
"would an emulator running on a 64-bit PPC be performance competitive with real 32-bit X86 hardware?"
Answer:
No, not by a long shot.
I also see comparisons of WebObjects to .NET, which really shows a lack of understanding about what .NET is. I don't pretend to understand it fully, but it is a lot more than the WebObjects are and it will have longer term and deeper meaning than any project Apple has. My guess is that Apple would do well to provide a MacOSX based .NET implementation.
I've got to hand it to you though, you're very verbose. You probably tend to win arguments by just out-writing the opposition, don't you?
<strong>I can't believe I read the whole thing (actually I didn't, I just scanned it). One question popped out at me though:
"would an emulator running on a 64-bit PPC be performance competitive with real 32-bit X86 hardware?"
Answer:
No, not by a long shot.
I also see comparisons of WebObjects to .NET, which really shows a lack of understanding about what .NET is. I don't pretend to understand it fully, but it is a lot more than the WebObjects are and it will have longer term and deeper meaning than any project Apple has. My guess is that Apple would do well to provide a MacOSX based .NET implementation.
I've got to hand it to you though, you're very verbose. You probably tend to win arguments by just out-writing the opposition, don't you? </strong><hr></blockquote>
On the first issue, that's why I asked it as a question. But can I just ask you to expand on why the gulf would be as large as you say. Surely Virtual PC isn't so bad on a well specced G4, or is it ???
On the second issue, I wasn't even vaguely comparing where Webobjects is now to where .Net is (BTW, I do understand what .NET does), but inferring that Apple needs to widen WebObjects to become a complete alphabet soup-compatible Web Services/EAI solution so that it has a competitive offering in that market.
The options here are: -
Don't evolve WO (and watch it get eroded) or
Evolve WO and/or
Get into bed with Sun's ONE initiative as a strategic priority.
The problem with application services is that a lot of people talk about them and few of them "get" the whole picture. My personal feeling is MS have been very clever in building a quite valid image of itself as a promoter of XML as a tool for "glueing" diverse applications together, and even more clever by presenting an image that it's the only game in town.
An alternate viewpoint would be that the strength of their offering is that MS has a de facto dominant position in so many marketplaces that it is able to deliver a complete portfolio of SOAP/UDDI/etc. enabled backoffice applications (SQL Server, Exchange Server, Content Management Server, IIS, etc. ad infinitum) that can be integrated using Visual Studio .NET either with VB, or C-sharp or whatever other languages are in there.
None of the other competitors in this market have the same breadth of products to deliver; Sun can deliver the iPlanet range for directory services. collaboration, and various others but has no database; Oracle has a database and a wide variety of horizontal apps, but no directory services; SAP and Siebel have suites of horizontal apps, with modules for specific vertical markets, but no directory services or own-brand database; Apple has very little other than a neat bit of middleware that can integrate with some iPlanet products and some public-domain elements like Apache.
So the reality is that .NET is a brand image which enacapsulates MS' XML/SOAP/WSDL/UDDI efforts, but it's the breadth of products behind it that give it substance and the importance that you quite rightly identify.
However, my position is that AAPL needs to either a) find some bedfellows it can trust to deliver (I am sharply reminded of the commercial fiasco that was OpenDoc [anyone reading from Novell, thanks for all your hard work there], which I personally thought was kind of neat) if it wants to counter .NET as an MS USP or b) it needs to invent like crazy, which again you rightly identify as a potential problem given the difference in size and resources.
On another subject, I don't win arguments by outwriting the opposition. But I do try to resolve them by discussing as many points-of-view as possible. I find that normally works better than just being dismissive of someone else's position. It's called debate.
On the first issue, that's why I asked it as a question. But can I just ask you to expand on why the gulf would be as large as you say. Surely Virtual PC isn't so bad on a well specced G4, or is it ???
<hr></blockquote></strong>
Processor emulation requires that each instruction in the original (emulated) program be matched with a series of native instructions which do the equivalent operations to an emulated processor state. This is because the original program was written assuming that it was modifing the state of the original processor, so the emulator has to reproduce that state and as it processes emulated instructions the actions of those are reflected on the "fake" processor state. Now add the overhead of getting the native (real) machine to decode the emulated instructions, as well as dealing with basic difference between the machines (byte-ordering being a major one). In the original 68040 emulator that Apple wrote when first switching to the PowerPC, it took about 10 PowerPC cycles for every 68040 cycle. This meant that a 150 MHz PowerPC was about the same speed as a 15 MHz 68040. Emulating an x86 won't be too far off this mark -- in some ways there is less machine state, but the instruction set is more convoluted and these days there are MMX and SSE units to deal with. The x86 is also much more memory intensive (due to the lack of registers), which doesn't really sit well with the PowerPC's load/store architecture. Also, the FPU on the x86 processors uses a floating point type that the PowerPC doesn't support (80-bits). As a result a 1 GHz PowerPC would approximate a 100 MHz x86.
One way to reduce this huge drop in performance is to identify sections of code that get executed repeatedly and convert them en-mass to a single series of "equivalent" PowerPC instructions. This works, but carries its own overhead of keeping track of when this kind of mass conversion should be done, whether it has been done, etc. Also, the emulator still needs to track the emulated machine's state. Apple's "dynamic recompliation emulator" for the 68040 reduced the performance from from about 10:1 to about 4:1. Thus a 1 GHz PowerPC would be equivalent to a 250 MHz x86. The current PentiumIV is almost ten times that fast.
Some of the performance issues can be mitigated by replacing key pieces of code with native PowerPC code that does the same thing. The biggest target is Windows or its drivers. SoftPC/VirtualPC implement some of the important pieces as real PowerPC code and thus get a real speed boost for software which relies heavily on Windows or drivers for its computationally intensive parts. Applications which bring their own computations to the party, however, are toast.
There are a bunch of other reasons why software written for one processor won't run terribly fast when emulated on another -- and software which is carefully tuned to eek out as much performance from the original processor as possible will suffer the most. That software also won't be able to take advantage of PowerPC techniques to maximize performance, so it'll suffer coming and going.
There are a couple of companies out there claiming that they have faster on-the-fly translation, but the demos have been pretty weak and their claims aren't really all that much better than this when you get right down to the nitty gritty. A 1.5 GHz can emulate a 1 GHz PowerPC when doing non-intensive stuff... oooh, wow. That's not where it really counts, is it?
[quote]<strong>
On the second issue, I wasn't even vaguely comparing where Webobjects is now to where .Net is (BTW, I do understand what .NET does)...</strong><hr></blockquote>
Okay. If Apple is going to compete in this arena, however, they should jump on the bandwagon that has the lead. As much as I hate to say it, Microsoft's song-and-dance is far more compelling than anyone else's. If Apple bought into it and did a MacOSX implementation (integrating it with their stuff, of course) that would both keep them busy and give them a fighting chance. This battle is about developer mindshare, and if Apple can be code-compatible then that is a huge leap forward for them. Beyond huge. I'm not sure about VB, but C# is a virtual machine like Java so a PowerPC native implementation could be competitive. Microsoft might just let them come along for the ride simply because that would be the straw that broke the camel's back.
Hmmm... could I fit another cliche in here? Probably not.
Okay. If Apple is going to compete in this arena, however, they should jump on the bandwagon that has the lead. As much as I hate to say it, Microsoft's song-and-dance is far more compelling than anyone else's. If Apple bought into it and did a MacOSX implementation (integrating it with their stuff, of course) that would both keep them busy and give them a fighting chance. This battle is about developer mindshare, and if Apple can be code-compatible then that is a huge leap forward for them. Beyond huge. I'm not sure about VB, but C# is a virtual machine like Java so a PowerPC native implementation could be competitive. Microsoft might just let them come along for the ride simply because that would be the straw that broke the camel's back.
Hmmm... could I fit another cliche in here? Probably not. </strong><hr></blockquote>
[quote] I don't object to you making an argument, but I think that they should at least vaguely observe some logic. <hr></blockquote>
[quote] HOS, a career in spin doctoring awaits you here in the UK should Tony Blair's Richlieuesque Communications Director choose to go and get a proper job. <hr></blockquote>
[quote] You're spinning again! <hr></blockquote>
<ad hominem attack>
Heh- the more you rely on ad hominem attacks, the less impressive your argument becomes... <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
</ad hominem attack>
Anyway, back into the fray:
[quote] bonus question, I believe its Monty Python (the argument sketch), but I cannot remember who the performers were. Palin & Cleese <hr></blockquote>
We have a winner! Although it's admittedly an easy lob since one would expect an Englishman to be aware of famous English comedy... <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
[quote] An example of what I'm complaining about would be: Apple currently makes mainstream systems using commoditized components which are sold at mainstream prices, therefore Apple cannot exploit the HPC computing market because they would have to retain a mainstream business model and could not therefore build a high-end system. <hr></blockquote>
Incorrect. My exact words were:
[quote] With prices between $800 and $4k, Macs are dirt-cheap... any HPC-like machine or technology that Apple employs (like a NUMA architecture) has to be cheap. Therefore, this must exclude many HPC characteristics which are too expensive to implement. <hr></blockquote>
Remember the context- this is discussing MCCFR's assumption that the Apple of today is working on or is interested in some sort of HPC or HPC-like technologies.
I stand by this statement, which is most definitely different than claiming Apple cannot ever build a high end HPC machine. I'm very clearly stating that to sell machines between $800 and $4k, Apple needs to use technology whose costs land under these magic figures. Whether Apple chooses to play in the $20k and up computing category is a completely different argument- and my arguments against that were stated in my later posts. You have my apologies if this distinction was not sufficiently clear.
Perhaps one or two more thoughts along this line to illustrate this.
Upcoming sub-$4k HPC-like technologies which could be used in Apple's $800-4k lineup include RapidIO, HyperTransport, 64-bit commodity CPUs (G5), RAID, and the kind of NUMA multiple CPU architecture discussed ad nauseum in the Dorsal M thread. These are the kind of commodity cheap HPC things I'm looking at- a high bandwidth switched bus, multiple CPU support (>2 anyway), large address space, commodity high speed networking, fast I/O.
Expensive HPC stuff which won't fit under $4k- massively parallel systems (>128 CPUs, heh, call it >4 to be more realistic), terabyte disk arrays, pervasive fiber interconnects, multi-terabyte robotic tape archival, N+1 redundancy for all components, automatic failover, pervasive massively wide busses with a direct crossbar topology between CPUs.
Thus, Apple can sell better performing machines for <$4k in the near future by borrowing some HPC concepts, independent of whether they choose (or choose not) to sell a $20k+ machine.
Continuing further, a past thread in my arguments is that this HPC-like technology _need not come from SGI_, lessening the perceived dependence Apple would have on SGI technology.
Next topic:
[quote]Using this argument, every marketplace that AAPL competes in is saturated <hr></blockquote>
Nope- an absurd abstraction ripped from context.
Apple has a dominant market share in more than one saturated market- desktop publishing comes to mind, and nonlinear semipro video editing is another. Apple has no market share in >$4k HPC, also a saturated market.
To ignore this distinction is a grievous error, as dominant market share imparts inertia. May I consider this self-evident, or should I expand on this?
[quote] The reason that companies run Sybase or Oracle on Sun Starfires is a relatively simple combination of factors: it's the only agressively marketed 64-bit general purpose business applications server with a comprehensive ISV business applications portfolio that runs a robust "open" operating system. <hr></blockquote>
This is directly ignoring my main point, which even though was quoted, I feel is so important, that I'll repeat it- most shops run Oracle on Sun, Dell, IBM, etc. hardware because Oracle has a proven, demonstrated track record of running on these platforms! Apple does not, thus giving it a significang hurdle it needs to overcome to compete here.
Re: Apps servers
[quote] I was not referring to narrow application servers, but to web application services where AAPL has WebObjects as a well-respected product. <hr></blockquote>
I don't have any real experience with WebObjects, so I'll be terse here. WO seems to be a "middleware" package which is largely restricted to being a method for developing web-based applications, in the same vein as openACS, ColdFusion, PHP, or ASP.
Based on my experience in the "New Economy", the current Xserve will do wonders here. It will scale through multiple Xservers, rather than needing a "big iron" Xserver.
Web apps in particular are usually network or DB constrained. Most are built in redundant configurations, so multiple devices are preferable to one big device. Again, this plays into current Xserver design.
OK- my lunch break is running short, and I have a conference call I can't get out of in about 20 minutes, so I'm going to have to cut things a bit short, even though I have a lot more to say.
[quote] I want to see AAPL leverage a scalable OS on a variety of system segments all the way from Home to Enterprise!<hr></blockquote>
Ah-hah, now the end goal is being laid bare! MCCFR's dream is to have Apple be a large Enterprise vendor (in addition to its current "creative ghetto"), a la HP or Dell.
This leads into what I was asking for in one of my earlier posts, about "Apple... by Apple" vs. "SGI... by Apple."
This argument relies on:
1) There exists this thing called brand.
2) Consumers have certain associations with brand
3) Brand has value (can be positive or negative)
4) Brand has cost- to build, to maintain, or to change
<strong>Uh-oh, things are beginning to turn ugly...
Ah-hah, now the end goal is being laid bare! MCCFR's dream is to have Apple be a large Enterprise vendor (in addition to its current "creative ghetto"), a la HP or Dell.
This leads into what I was asking for in one of my earlier posts, about "Apple... by Apple" vs. "SGI... by Apple."
This argument relies on:
1) There exists this thing called brand.
2) Consumers have certain associations with brand
3) Brand has value (can be positive or negative)
4) Brand has cost- to build, to maintain, or to change
More later, got to run.
-HOS</strong><hr></blockquote>
Firstly, I apologise for the perceived attacks - my intention was to point out that your version of "truth" is not necessarily mine or anyone elses. Likewise, the "possibility" that I'm expounding may seem like the ramblings of an idiot to some and the very crystallisation of truth and beauty to others.
We all put a spin on facts so that they support our point of view, I was just being clear in saying that was what I felt you were doing and - by comparing you to Alistair Campbell (Blair's Eminence Grise) - I was pointing out the quality with which you were executing said spin.
Also, my dream is not to have AAPL as a Large Enterprise Vendor as you put it. My dream is to have AAPL as a definitive vendor of computing solutions, no matter what the market context (Home/SoHo/SME/Enterprise/Government). Personally, I do not feel that anyone has yet rebutted this as an aspiration.
On another note: Yes, you can run 20 xServe models to deliver a web applications farm. But in a commercial world where consolidation is a pressing requirement, maybe larger businesses would like to be able to run 10-20 virtual servers on a single larger box. If you don't believe me, look at vmware.com, or join me on a project whose aim is rationalise server count on a 3-1 or 5-1 basis.
But the things that really get me about many of the arguments that the naysayers to this thread have put forward are as follows:-
The total denial that this could even be a possibility
The abject lack of ambition for AAPL to extend itself into the corporate market
The lack of confidence that AAPL could even attempt to execute such a shift in its aspirations, let alone succeed in the attempt
I've constantly said that I doubt the possibility myself: but the difference is that, despite the doubt, I was willing to explore the opportunity. Maybe Raycer or nVidia or whatever may be some part of AAPL's future. But that does not necessarily extrapolate to it being ALL of AAPL's future.
First of all I would like to congratulate everyone in this thread for contributing to what has to be the first intelligent discussion I have read on an Apple-centric forum. To have a thread such as this not descend into ignorant and misinformed Microsoft bashing, or tears and recriminations about when Apple is going to drag its core technologies into the late twentieth century never mind the early twenty first, is refreshing enough for me to contribute to my first Apple forum thread ever.
I want to return to this assertion:
Apple plus (purchased) Sgi plus (renewed co-operation with) Avid = Total domination of film and television (post-production)
Note that I have expanded this somewhat.
Firstly, let me explain my own career path, so as to establish some credentials:
Graphic Designer (on Macs when 33mhz and 16mb of ram was to die for)
to
Broadcast Designer (using Quantel Hal,Harriet and Harry, highly specialised and proprietry systems)
to
Avid Editor (Avids running on both Macs and NT)
to
Flame/Fire Operator (Discreet compositing and editing systems on SGI Octane and Onyx)
to
Post Production facility Sys Admin (running Discreet on Sgi, Avid, Shake, Maya etc on Win2k)
With Apple's recent purchase of Nothing Real and Silicon Grail, it would seem to me that they are making a fairly concerted and frankly unwelcome move into my business. I should state that the market we cater for is somewhat higher up the scale than Final Cut Pro editing DV. Think commercials with eye-watering budgets and ridiculous deadlines. My initial reaction has been to dismiss any further use of Shake and Rayz past the versions my facility is presently running (no big deal, many post houses still use Cineon, which became the core of Rayz) due to the fact that I would not even consider having a Mac in the facility (other than as part of an Avid, and even then I'd probably go with Win2k). They simply are not fast enough for what I need them to do in the time I need it done in, and I'm sure Apple is as aware of that fact as I am.
However, Apple purchasing and leveraging the technology of SGI puts me in two minds. Firstly, it would be reassuring for me for SGI to be in something of a more secure position. SGI today is Apple in the mid-nineties - fantastic technology, dismal company. Apple today is the reverse as far as I am concerned - there is only so far you can take ridiculosly lacklustre hardware in pretty boxes.
The reverse of that, however, is nagging doubt over Apple's focus. This the company that has chased market sectors over the last decade like the drunken party girl at a fraternity party, waking up in the morning with a headache, someone in bed next to them they don't recognise and no money in their wallet. Who is to say that my industry isn't simply Apple's latest night on the tiles, to be dropped when they sober up.
To return to the core of this thread, I can see incredible synergies between SGI and Apple, lets face it, Apple needs someone to teach them how to write a proper Unix for a start, by that I mean one that runs at the speed a Unix core on a RISC chip should reach. Both systems have always been known as the artist's computer. Both deal with niche markets. It's a shoe-in, and I can see it happening.
i stopped posting because this thread had gotten into economics and accounting practices, not my territory.
shannyla, that was a good post, sums up my thoughts on the subject. your post is a good representation of what i was saying about perceptions in the high end market, ive heard from allot of rayz users that they wont be migrating to the mac, too bad for apple. what will they do to win market segment, buy sgi perhaps. i still contend that its a great idea, cause it will make my life easier, and who dosn't want that.
Apple needs someone to teach them how to write a proper Unix for a start, by that I mean one that runs at the speed a Unix core on a RISC chip should reach.</strong><hr></blockquote>
...that the speed (or lack of...) of MOSX is making them and their applications look bad, then I would suggest that the OS is in serious need of some fettling.
The whole point of Unix OSs is that they take up tiny amounts of resources, which is why Irix is so rapid on systems with relatively low Mhz counts. One of the problems with MOSX, from my and my industry's point of view, is that is full of computationally demanding junk. Pretty interfaces and bouncing dock icons are fine and dandy for making the sale to Johnny at CompUSA, but they don't mean shit to me. The same goes for other demanding application areas such as Medical Imaging, Industrial and Architectural Visualisation and Biotech.
Now this wouldn't matter a damn to me, but Apple decideded to come crashing into my area, in a frankly gauche and careless manner. So there are reports that Pixar have swapped to Macs, I very much doubt it except for a few high-profile machines for cross-marketing purposes. Where are the high-end OpenGL cards that Maya needs? (If you say Geforce4, you are an idiot and we both know it...) Where is the truly powerful render-farm solution? Xserve isn't it, that is the most fundamentally flawed server I have ever seen. Where is the redundancy? Where is the SCSI? I wouldn't trust my mission-critical systems to IDE drives with tiny MTBFs. Add to that the requirement to rewrite everything, and I would suggest that all that has happened is that Pixar has locked the door to their Sun machine room and put up a sign saying "there's nothing here for you, move on now"
Comments
<strong>OK, a little bit more-
MCCFR, you have several strong arguments. Since I'm too lazy to go look up SEC filings on SGI and crunch my own numbers (I've not done any diligence at all, due or otherwise on SGI since they bought out Cray, and I had a college professor assure me that "this means supercomputers on your desktop in five years!"), so I'm definitely arguing from a weaker perspective. Especially since my position is that costs are too great for value received.
So instead of doing real work, I'll take the lazy way out and go after the assumptions- if I can topple enough of them, then the subsequent conclusions based on them will topple as well.</strong><hr></blockquote>
OK, I don't mind the debate.
[quote]<strong>Assumption 1) Apple is working on a HPC product, or a product with HPC characteristics.
HPC is a very broad field, and traditionally ranges from individual vector processors (older Crays) to massively parallel machines (Beowulf clusters, ASCI White), and encompasses many things inbetween (Sun E15k). Prices are accordingly high- often starting in the $100Ks and scaling up to the millions.
There's no doubt in anyone's mind that Apple is interested in building computers which have better performance than their current machines. But a little bit better and orders of magnitude better both qualify as better than current offerings.
With prices between $800 and $4k, Macs are dirt-cheap compared to most all of SGI's offerings. In order to maintain these low prices, Apple adopts far more pedestrian, commodity parts.
Rather like the old MG or Triumph approach to performance- recycle existing parts in an exciting shell. (OK- cheap shot, but my last British car was my Dad's '52 MGTD... which doesn't go from 0-60 mph as it tops out about 55...)
Back to the argument, any HPC-like machine or technology that Apple employs (like a NUMA architecture) has to be cheap. Therefore, this must exclude many HPC characteristics which are too expensive to implement.
So if Apple is looking at HPC, it's looking at something really small-scale or cheap HPC, so we have minimal overlap with what SGI does... or charges.</strong><hr></blockquote>
A significant element of this statement is logically constructed, however you are omitting information or excluding certain concepts that for your own purposes.
I would have imagined that - if Apple had an interest in SGI - it would be to exploit the foundations that SGI have built. So cc:NUMA would be one of those cornerstones.
The challenge to Apple is not to produce a machine that is cheap, it is to produce a solution which scales both technically and financially and can deliver ROI through competitive differential advantage. I would argue that one of the SGI Origins (300, though it might be 3000) is the prototype for such an approach.
I am proposing that whilst SGI has an innovative HPC line, they are, becuase of their marketing and distribution model, unable to exploit anything more than a nominal market share.
Apple has the brains and the money to adapt this headstart so that new, innovative server platforms can be developed in a variety of product classes.
Small Business - General Corporate
=========================
AAPL have already released xServe, which competes against Compaq's DL380G2 in terms of spec and price.
From a technology perspective, this unit is probably where it needs to be: The addition of a "G5", with true DDR (hopefully at 400MHz+ running a 32-bit OS XI would just about be Nirvana.
Medium Business - General Corporate
===========================
I would have assumed that this would be AAPL's next server, released just after MWSF 03.
More than a simple "bigger" server, AAPL must release this to service 250-500 regular concurrent users. My guess would be a system that can go from 2 to 4-way as standard, and with a true SCSI (six drives) array, internal expansion bay for DLT/AIT, plus six HyperTransport/PCI-X compatible expansion slots.
AAPL would be pitching this machine into the ML370G2 class, for file/print, but with the advantage of being scalable to 4-way and a 64-bit OS could easily pitch against IBM's x255/X360 (not 64-bit, but Xeon MP which given Itanium's performance is no bad thing).
Large Business - General Corporate
===========================
Joint-top Billing with G5 TiBook (MWSF 04).
Keyword: Fabric
Key Technology: InfiniBand
A modular server product comprising up to 4 core modules (each delivering either 4-way or 8-way processor modules) with the ability to deliver 64GB in RAM and six HyperTransport slots in each core module, additionally: -
Core storage module (4 x SCSI-320 drives)
Sharable (two host) expanded storage module (2 x 7-slot , deploying up to 14 x 240GB+ fibre channel native drives (i.e. 3.2TB formatted) or 7 x 240GB + 1 x tape module (DLT/AIT) or 1 x Tape Changer.
Sharable (two host) expansion module delivering 12 HyperTransport/PCI-X or slots or better.
Target Market : Database + Apps Server + Small Scientific/Visualisation
Current key competitor: IBM X440 (again, Xeon MP and 32-bit only)
All I can say so far is, where do sign !
But this is where my plan gets interesting.
Large Business - Scientific & Visualisation
===========================
Time: SIGGRAPH 04 or MWSF 05 or NAB 05
Target Market : Data Warehouse + Apps Server + Large Scientific/Visualisation
First thing up is Visualisation Module, called something squiffy like Infinite Reality, containing the latest evolution of all of SGI's visualisation technology. Also delivers optional digital video I/O interface, such as SDI and Gigawire. (SGI guys front and centre). Fits into Infiniband-fabric server announced at MWSF 04.
Announced to cheers from the multitude, SJ shows off the first AAPL implementation of SGI's Reality Center, with an immersive VR screen the width of the Moscone Center stage, SJ braves the attack of several dozen Empire fighters before destroying the Death Star.
But surely that wasn't all done with just 32 G5 processors
SJ now announces one last thing: A router box for the management of a single coherent shared memory which connects to an implementation a 12X 30Gbit (3.75GB) InfiniBand interface.
A few bits of software jiggery-pokery (SGI's software engineers front and centre) and up to 64 8-way modules (which take up no more than 3U of rack space each) can be connected together to deliver a 512-way MPP system that delivers 4TB of coherent shared memory (so normal programmers can use it easily).
Allowing for a 5U router module for each rack, plus UPS and other ancillaries, Apple would have a solution which could scale from a single 22U rack (servicing several hundred database users or a hundred visualisation users or a couple of thousand broadband streaming customers) to (dependent on storage and tape backup requirements) 10 or more 42U racks (servicing several thousand data warehouse users or several hundred visualisation users).
This is the core of my server argument: AAPL can evolve and scale a single highly marketable product (4-way, 64GB, 6 PCI-X) with an enormous audience into a highly glamorous, highly profitable, HIGH PROFILE product which will make Sun, IBM, and HP/Compaq look around desperately to create some kind of viable response.
In terms of density, protected investment, consistency of support , this would be a world-beating range.
[quote]<strong>Assumption 2) SGI is working on a PPC-based machine. Not provable one way or the other, and if not true, dampens a lot of the fire since this assumption is crucial to transition costs. If true though, this machine is more likely to be a million dollar server than a $1k blade. Again, we're not guaranteed that costs of this technology fall into line with what Apple can afford to charge, or even be the kind of technology that Apple's really interested in at the moment.</strong><hr></blockquote>
I have to say this argument ignores commercial reality: when VW purchased Bentley, did they say "We'd better be careful, because VW can only afford to charge $20K to sell to its traditional audience" ?. I think it is far more likely that they said:-
[quote]<strong>"Here is a company whose brand image and values will be commercially exploited by either DaimlerChrysler or BMW if they get hold of it. We can afford to make Vickers an offfer and enhance our image in the marketplace, which will hopefully lead to more sales for our other brands. Oh, and they sell for a fortune to a loyal audience with more money than sense, so we can make a profit!".</strong><hr></blockquote>
Replace DC and BMW with Sun or Compaq, replace Vickers with "SGI's shareholders" and I would have thought that was a position that Apple's board could take.
As for the G5 thing! Have I not been clear enough on this point - it's a rumour, that's why I occaisonally use the phrase "If The Rumour is true...".
It probably is not true, but then again it might be given the MIPS market positioning thing I mentioned a few posts back.
Maybe SGI have taken the view that MIPS are going in a direction (embedded, mobile, games console) that is incompatible with the mainstream markets that SGI have to pursue, they need to align themselves with another high-performance processor design which has a long-term future. I don't know, it could all be the product of an overactive imagination which was never apparent in my school years.
[quote]<strong>Assumption 3) SGI is sufficiently inexpensive that Apple can pick it up without too much sweat. MCCFR has done a very good analysis demonstrating that this is probably true. The only real hole I can knock in here is that even though it's within Apple's budget, it still represents a sizable expenditure which may be better spent elsewhere.</strong><hr></blockquote>
Might be!, Might not be! If my aunt had a moustache, she would have been my uncle, Hold on, my aunt does have a moustache, but I know she's not my uncle.
I think all of these questions can be solved with due dilligence, although - especially with my aunt - I'm glad I'm not responsible for any part of it.
[quote]<strong>Assumption 4) Synergy in Apple's and SGI's video editing and 3D graphics markets.
While true to a degree, key apps like Maya already exist for MOSX, so other than control over the direction of Maya, the major benefit of Maya-on-the-Mac already exists. More importantly, the key infrastructure of generating video editing software (iMovie, Final Cut Pro) is already in place and running on both today's and yesterday's Macs. Whether software like this would run so much faster on future SGI hardware or future Apple hardware as to justify the cost of making it happen is up in the air. I don't know if high-end markets like television stations are willing to fork over the $100k for a high-end Final Cut Pro on SGI-class hardware, or whether $4k Apple hardware is sufficient to get the desired work done. The one thing I can easily state is that the demand curve clearly shows that more $4k boxes will sell than $100k boxes.</strong><hr></blockquote>
The art of a good Scrabble player is someone who can take another player's word, add a few letters, taking in a double-word score as they go.
So, "The major benefit of Maya-on-the-Mac exists" becomes "The major benefit of Maya-on-the-Mac exists for as long as SGI continues to be independent of Microsoft or another predator".
Equally, "I don't know if high-end markets like television stations are willing to fork over the $100k for a high-end Final Cut Pro on SGI-class hardware", becomes "I don't know if high-end markets like television stations are willing to fork over the $100k for a high-end Final Cut Pro on SGI-class hardware, but they might handover $100K if they got online compositing, plus editing, plus whatever else".
As for "Whether $4k Apple hardware is sufficient to get the desired work done", I'm not sure I'm offering a $4K solution, but I think I might be offering a powerful $20K+ solution.
BTW, can someone verify this equation: (AAPL*QT6)+ SGI + AVID = market dominance of professional and corporate audio production + broadcast news/film/video/DVD post-production + class-leading broadband streaming of the end product.
Yours for a $750M combination of cash + shares, plus restructuring expenses and ongoing investment.
[ 06-18-2002: Message edited by: Mark- Card Carrying FanaticRealist ]</p>
Re: Apple and HPC
[quote]
A significant element of this statement is logically constructed, however you are omitting information or excluding certain concepts that for your own purposes.
<hr></blockquote>
Yup- an argument is an intellectual process, not the automatic gainsaying of anything the other person says. (Bonus points for whoever can name where that line came from!) Besides, isn't one of the goals of rhetoric to not destroy one's argument by contradicting it, something I admit I'm not always very good about?
[quote]
Small... Medium... Large Business
<hr></blockquote>
One of the problems with this is that it presupposes Apple's interest in "climbing up the food chain" to play in Medium and Large Business computing. Xserve is pretty limited in its target market (small server), and worse, the Medium and Large market is largely saturated as is with competitors like IBM, Dell, Hewlett Compaqard and Sun. Further, the technology Apple is using or could use as proposed (with SGI buyout) is not sufficiently compelling to grab a significant chunk of Medium and Large IT budgets.
I don't think that it's a stretch to say that Apple has stayed out of these markets since the return of Jobs because Apple doesn't want to play in these markets, instead concentrating heavily on the consumer and desktop publishing markets. Now, of course, they're trying to expand the "creative" market by pushing into video editing. As an aside, 3D modeling seems to be coming along for the ride, but doesn't appear to be a major focus of current Apple strategy.
If you look at the targeted markets mentioned above-
[quote]
Target Market : Database + Apps Server + Small Scientific/Visualisation
<hr></blockquote>
Most of these don't line up well with Apple's strengths. Database in particular- more people run Sybase on Suns for a reason. Oracle 9i is just now being ported to MOSX, and I'll go out on a limb here and say that the people currently buying Suns, Dells, IBMs, et. al. to run Oracle on will keep buying Suns, Dells, IBMs, et. al. because for at least the near term, Oracle has a proven track record on them. These vendors have demonstrated DB quality that Apple will never have until they've shipped DB quality machines long enough to demonstrate that quality.
Apps servers are all environment and application specific. Again, Apple is late to this party, and so lacks any first-mover advantage.
Scientific and Visualization here has at least some overlap with Apple's vision (BLAST for example), but one of the big trends here is for a distributed computing model- one where the XServe has as many advantages as the competition.
Expending effort to "crack" these markets will simply drain Apple's coffers and distract from the areas Apple wants to compete in- consumer, education, and creative markets.
[quote]This is the core of my server argument: AAPL can evolve and scale a single highly marketable product (4-way, 64GB, 6 PCI-X) with an enormous audience into a highly glamorous, highly profitable, HIGH PROFILE product which will make Sun, IBM, and HP/Compaq look around desperately to create some kind of viable response. <hr></blockquote>
I don't think this is very realistic. As I mentioned before, this is a very crowded market to debut a product like this into. One might even say saturated. Chances of success are pretty slim without some major "killer app" to drive it, be that price, performance, or software-based.
Re: SGI working on a PPC-based machine:
[quote]I have to say this argument ignores commercial reality: when VW purchased Bentley... <hr></blockquote>
I disagree. I think that's not a very good analogy for many reasons, not the least of which is the fact that VW is already composed of several different brands, something Apple is not. So adding a new brand is a different concept than buying out and shutting down a different company to leverage yourself into a new-to-you market.
Conversely, looking back at other arguments, your proposed strategy is more one of "SGI... by Apple" instead of "Apple... by Apple." So perhaps what you're really proposing is melding SGI and Apple together to create a different company than either is today. Whether this corresponds to the direction that Apple is currently headed in or not is an interesting question. MCCFR, if you choose to respond to this post, could you please expand a bit more in this line? Thanks in advance...
[quote] As for the G5 thing! <hr></blockquote>
Yeah- it's an Emeril Live! thing (Bam!), the argument is kicked up a notch if the rumor is true. If not, then SGI just looks more overpriced. Regardless, it's a key assumption in the cost equation.
Re: SGI is inexpensive...
[quote]If my aunt had a moustache <hr></blockquote>
Sorry to hear about your aunt. Personally, I've decided not to make fun of hairy people anymore because the older I get, the more (and in odder places!) hair I get. It's one of those "glass houses" things...
Re: Synergy...
[quote] The major benefit of Maya-on-the-Mac exists for as long as SGI continues to be independent of Microsoft or another predator <hr></blockquote>
In context, this actually isn't that big of a problem. One key assumption, that A|W is independent of M$ or another predator is self-evident by the fact that, despite the "bargain basement price", no one wants to buy out SGI and "Chainsaw Al" it into its components. In other words, SGI isn't in play, and it isn't because no one wants to spend $500-750M in cash or equivalents to do it. Which clearly shows that A|W isn't worth $500-750M plus the trouble of chopping it out of SGI. So that's a natural defense against it getting bought out.
Even more, small companies that are under control of "predators" still produce software for the Mac market... because there is money to be made doing so. Avid, for example, despite attempting to leave the Mac market, still makes hardware and software products for it. Avid very much shot itself in the foot for this, and is not in very good shape for it. Rather a lesson for anyone who would want to buy out A|W and shut down a very popular platform for its premiere (heh! pun intended) product...
-HOS
<strong>This is the most wordy thread I have ever witnessed. I think i just wasted about an hour reading idle ramblings and such. Keep it up.</strong><hr></blockquote>
Well, at least that's one thing this thread has achieved! <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
As for HOS, will respond as soon as I've had a chance to consider your arguments, except on one point on which I agree.
What I object to about middle-age is the way the hair migrates from your head, to your ears and nose! If there is anything more annoying than one of those thick, short, stubble hairs rooted in your earlobe, I have yet to find it.
I admit this is not about Future Hardware, but I would buy any gadget that dealt with this issue - Remington clippers don't do the job, cos the hair is too short.
On the bonus question, I believe its Monty Python (the argument sketch), but I cannot remember who the performers were. Palin & Cleese, perhaps or Palin & Jones. maybe someone could remind me.
I don't object to you making an argument, but I think that they should at least vaguely observe some logic.
An example of what I'm complaining about would be: Apple currently makes mainstream systems using commoditized components which are sold at mainstream prices, therefore Apple cannot exploit the HPC computing market because they would have to retain a mainstream business model and could not therefore build a high-end system.
This, to be blunt, is abject nonsense of the most absurd kind. Nothing is written in AAPL's articles of association that dictate that change either by way of market segmentation or product & market diversification is forbidden. So the only thing that prevents AAPL from improving its brand image in the corporate, scientific and specialist user markets is the will to do it, whether from scratch or by a process of acquisition.
[quote]<strong>
------------------------------------------------------------------------
Small... Medium... Large Business
------------------------------------------------------------------------
One of the problems with this is that it presupposes Apple's interest in "climbing up the food chain" to play in Medium and Large Business computing. Xserve is pretty limited in its target market (small server), and worse, the Medium and Large market is largely saturated as is with competitors like IBM, Dell, Hewlett Compaqard and Sun. Further, the technology Apple is using or could use as proposed (with SGI buyout) is not sufficiently compelling to grab a significant chunk of Medium and Large IT budgets.</strong><hr></blockquote>
Using this argument, every marketplace that AAPL competes in is saturated with everything from Tier 1 suppliers like IBM, HP/CPQ and Dell through mail-order to a million Main St. PC shops with beige boxes to sell.
As for making a compelling product, I would dispute that aggressively and will do so throughout this post.
[quote]<strong>
I don't think that it's a stretch to say that Apple has stayed out of these markets since the return of Jobs because Apple doesn't want to play in these markets, instead concentrating heavily on the consumer and desktop publishing markets. Now, of course, they're trying to expand the "creative" market by pushing into video editing. As an aside, 3D modeling seems to be coming along for the ride, but doesn't appear to be a major focus of current Apple strategy.</strong><hr></blockquote>
HOS, a career in spin doctoring awaits you here in the UK should Tony Blair's Richlieuesque Communications Director choose to go and get a proper job.
From personal experience, I would say that the reason for AAPL staying out of those markets is not that it chooses to, but that its invitation to the right parties "got lost in the post" because it didn't take the "privilege" of working with major corporates seriously.
I have a shameful secret here: Having worked with a major corporate in the oil sector under contract delivering Mac projects and support for three years from 1992, I was ultimately involved in migrating several key elements of its business away from Mac to Win95/O95 (client) and NT 3.51/4.0 (server). Why? Because AAPL in the UK didn't take the threat of an aggressive approach from MS to the customer seriously, and AAPL at a global corporate level (the era of Sculley and Spindler) didn't take charge of the situation.
That's as much information as I can share on a public forum, but you can read what you want between the lines and its probably true.
Working with major corporates doesn't necessarily make you the profit you want, but it does keep your profile afloat for smaller corporates and SMEs who like to go with the flow.
[quote]<strong>
------------------------------------------------------------------------
Target Market : Database + Apps Server + Small Scientific/Visualisation
------------------------------------------------------------------------
Most of these don't line up well with Apple's strengths. Database in particular- more people run Sybase on Suns for a reason. Oracle 9i is just now being ported to MOSX, and I'll go out on a limb here and say that the people currently buying Suns, Dells, IBMs, et. al. to run Oracle on will keep buying Suns, Dells, IBMs, et. al. because for at least the near term, Oracle has a proven track record on them. These vendors have demonstrated DB quality that Apple will never have until they've shipped DB quality machines long enough to demonstrate that quality.</strong><hr></blockquote>
The reason that companies run Sybase or Oracle on Sun Starfires is a relatively simple combination of factors: it's the only agressively marketed 64-bit general purpose business applications server with a comprehensive ISV business applications portfolio that runs a robust "open" operating system.
If AAPL delivered a modular server platfom, scalable from 2-512 processors, and with any visualisation technology made modular and, by implication, optional, it would have a solution that was (at worst) highly competitive in the marketplace or (at best) a definitive technology platform which could outscale and outperform any platform in its class.
Performance, combined with strategic applications like 9i or Adaptive Server (which is also being ported to OS X) and supportive/assertive developer relations marketing will lead to ISV's who have not been traditionally associated with Apple - like SAP, Siebel or Documentum - porting their applications to OS X.
AAPL would also have the advantage of being able to deliver a viable range of general business desktop/laptop systems which contributes to a lower TCO argument against Sun (and other Unix platforms) and counters the same argument from the WinXXX market.
If you accept all of these aspirations, then AAPL would then be ideally placed to deliver consistent 64-bit performance in all major sectors (client, workgroup, server room and datacenter) and a consistent user experience across all of those segments. This last point is more important that might first appear, but I feel the justification belongs later in this post.
[quote]<strong>Apps servers are all environment and application specific. Again, Apple is late to this party, and so lacks any first-mover advantage.</strong><hr></blockquote>
I was not referring to narrow application servers, but to web application services where AAPL has WebObjects as a well-respected product. However that respect does not necessarily translate into sales: This, I suspect, is partially because AAPL is not seen as being aligned with the mainstream industry, which a broader product range and a more comprehensive ISV business applications portfolio would undoubtedly address.
In addition, AAPL could/should try to broaden the functionality and appeal of WebObjects, so that it is more capable of answering the needs of the wider application services market, thus providing a viable alternative to .NET which MS will probably use as a sledgehammer to crush any development language that is not developed at MS, such as Java or Objective-C. Partnership with MS here is not practical as it would lead to junior partner status for AAPL, but marketing WebObjects as an "open" alternative to .NET will give dissentors a mainstream place to go.
[quote]<strong>
Scientific and Visualization here has at least some overlap with Apple's vision (BLAST for example), but one of the big trends here is for a distributed computing model- one where the XServe has as many advantages as the competition.</strong><hr></blockquote>
This may be true in some areas, but it is not a universal truth. Speaking from my most recent flirtation with the Oil/Petrol industry, I noticed that there were more Unix servers (of all complexions) in the primary server suite than the visit before that. I can't help but feel that commercial science will always have a tendency to gravitate to centralised computing because of support arrangements.
[quote]<strong>Expending effort to "crack" these markets will simply drain Apple's coffers and distract from the areas Apple wants to compete in- consumer, education, and creative markets.</strong><hr></blockquote>
You're spinning again! Wants to compete in or has to compete in ? I feel you're turning a necessity into a virtue. Quite frankly, this is the equivalent of a car manufacturer like Rover saying it's not interested in the North American marketplace, whereas the reality of the situation is that it got its butt severely kicked around the last time it tried so it went off with its tail between its corporate legs.
To paraphrase Oscar Wilde, AAPL so far has done a fair job of convincing two audiences who "know the cost of everything" i.e. consumers and education. But in the former case, there is still a lot of work to do to increase user base (whether expressed as market share or absolute numbers) and the latter case is dependent on the political priorities of a given state or school district.
In addition, both of these markets are dependent on the health of the economy: the former is dependent on discretionary expenditure and the latter is dependent on a vibrant tax base.
Another issue is that AAPL's success in the K-12 market is not universal: In Europe, much of that business goes into the WinXXX market, because (rightly or wrongly) that is seen as an affordable option for students in higher education and seen as the de facto standard in business.
I concede that AAPL will always see the "creative markets" as its home turf. However, I am sceptical that we can always rely on that market as a cash cow.
The laughter that used to accompany discussions of Windows 3.x or Windows 95 being used for serious creative work has become a little more nervous in the face of WinXP (Fisher-Price toy interface or not) and the perceived performance gap that now exists.
The fact that the New Media industries have never bought the "Macintosh Uber Alles" party line to the same extent as print media is telling. The desire for colour management is not quite as extreme in the web world and New Media tools evolved in the WinXXX universe more quickly than on Mac resulting in an obvious bias towards WinXXX.
The market that remains is that for A/V post-production and workflow, which as you rightly point out is the market that AAPL are obviously lining up for dominance given the recent acquisitions. And I genuinely hope they succeed!
However, I believe that the time has come for AAPL to try and stretch beyond the status of a "creative ghetto" system or a system for domestic consumers with an eye for design flair.
I want to see AAPL leverage a scalable OS on a variety of system segments all the way from Home to Enterprise!
I want to see AAPL create a set of USPs for itself, which allow customers (both existing and potential) the opportunity to create USPs for their businesses!
Where there is a business or scientific challenge, I want to see AAPL being considered as a potential solution to that challenge!
I am passionate about this product and its future, and I believe that the future for AAPL is to get back into the corporate market.
[quote]<strong>I don't think this is very realistic. As I mentioned before, this is a very crowded market to debut a product like this into. One might even say saturated. Chances of success are pretty slim without some major "killer app" to drive it, be that price, performance, or software-based.</strong><hr></blockquote>
First off, I don't think we should describe price or performance as "killer apps", otherwise the whole benchmark or MHz thing becomes the killer argument whereas it should be an equal factor in the evaluation process.
Defining a killer app is difficult, because that app varies from industry to industry. But let's explore a few possibilities which I will ask as questions because I'd like to hear a few different opinions.
Solution A
======
Looking at server consolidation, which is a vogue at this time, could AAPL use the skills of a Connectix and/or VMWare to develop a partitioning solution for their servers so that a single 32-way server could be utilised to run Win2K on two, three or four partitions (let's say two x 1-way, two x 2-way). 64-bit OS XI could be run on 2 x 8-way partitions 1 x 4-way partition and 1 x 2-way, and Linux could run on the remaining 8-way partition.
Would Win2K (or it successors) running under emulation on a 64-bit platform be performance-competitive when compared with running native under Intel or AMD 32-bit architectures? Can someone answer that?
I'm not wholly keen on this as a "killer app", largely because it feels uncomfortable at a religious level giving MS another way to implement its "solutions". However, it would recognise the fact that many customers are committed to using using MS for Exchange and AAPL do not have an enterprise e-mail system that they can put up as a solution.
Solution B
======
Another popular corporate technology (certainly since Win2K) is Citrix MetaFrame, which - for those who are unaware - is a solution for delivering thin-client Windows sessions either over a) slow WAN links or b) on clients which no longer meet minimum equipment specs to run Win2K or c) for corporate apps which are considered to be appropriate for centralisation which is deemed to be a cost-saver.
In reality, MetaFrame takes a significant influence from windowing technologies such as X/Windows, where I think I'm correct in saying that presentation was the preserve of the "client" system but computation was the preserve of the "server".
The problem with MetaFrame is that it is quite stunningly expensive to implement (both in terms of license fees and the quality and depth of infrastructure that is required to run the solution).
So my question is: Could AAPL adapt an evolution of X/Windows (Unix core after all), so that personal productivity apps (such as Office) run as purely client apps, but business management apps (e.g. SAP) could run as windowed apps? And if this were technically feasible, could it be delivered with typical Apple flair (as opposed to MetaFrame, which is not as pretty as it could be) and at a cost which significantly undercut MetaFrame?
If you combined A and B with cost-effective SMP or HPC systems, AAPL would surely have a set of tools which constituted "bait" for corporate IT functions to wean themselves off Intel-based infrastructures.
If you accepted all of my arguments in this post, then really the only major tool that AAPL would be missing from its arsenal to establish a comprehensive range of IT solutions at the end of this process would be an Exchange-killer. I have to admit on this point, I have no opinion or way forward - my opinion of Notes is pretty low and my opinion of Outlook (on Windows) relatively high, so I would appreciate some input on how that issue could be resolved.
My own thought was whether Sun would ever consider making some of the iPlanet products available under OS X, but I have no knowledge of these at all (except evaluating the product that became the collaboration server when it was Collabra) so would appreciate some input.
[quote]<strong>I disagree. I think that's not a very good analogy for many reasons, not the least of which is the fact that VW is already composed of several different brands, something Apple is not. So adding a new brand is a different concept than buying out and shutting down a different company to leverage yourself into a new-to-you market.</strong><hr></blockquote>
Ok, but your original argument which caused me to use that counter was something like: Could Apple sell outside of its traditional marketplace?
My answer to that is still "Yes", and if you dispute my first analogy than I will simply have to use VW again and point you at that company's soon-to-be-released Phaeton. Whether a company (whose name translates as "people's car") selling a model for £60K+ is ironic, I will leave for others to discuss. But it is outside of VW's standard market.
[quote]<strong>
Conversely, looking back at other arguments, your proposed strategy is more one of "SGI... by Apple" instead of "Apple... by Apple." So perhaps what you're really proposing is melding SGI and Apple together to create a different company than either is today. Whether this corresponds to the direction that Apple is currently headed in or not is an interesting question. MCCFR, if you choose to respond to this post, could you please expand a bit more in this line? Thanks in advance...
</strong><hr></blockquote>
I'm not a marketing guru, so I'm a little unsure of the best way to proceed here, but I'll give it a bash.
I think the relationship that I would choose for SGI would be analogous to the relationship that AMG has with Mercedes-Benz or Alpina has with BMW. I believe that the core technologies should be marketed as an Apple solution, but wherever SGI technologies have been deployed to enhance the core solution, that the enhancement is recognised along the lines of "Apple, tuned/enhanced/whatever by Silicon Graphics" .
This would stress the foundation of the system, but also continue the exploitation of SGI brand and provide continuity for customers and employees alike.
Does this answer your question?
[quote]<strong>Yeah- it's an Emeril Live! thing (Bam!), the argument is kicked up a notch if the rumor is true. If not, then SGI just looks more overpriced. Regardless, it's a key assumption in the cost equation.</strong><hr></blockquote>
OK, let's accept that as a given.
[quote]<strong>In context, this actually isn't that big of a problem. One key assumption, that A|W is independent of M$ or another predator is self-evident by the fact that, despite the "bargain basement price", no one wants to buy out SGI and "Chainsaw Al" it into its components. In other words, SGI isn't in play, and it isn't because no one wants to spend $500-750M in cash or equivalents to do it. Which clearly shows that A|W isn't worth $500-750M plus the trouble of chopping it out of SGI. So that's a natural defense against it getting bought out.</strong><hr></blockquote>
And I accept that for the moment, but there may come a point where SGI is an animal so weakened by fighting its corner that the opportunity for predators becomes irresistable. My worry is what happens thereafter.
[quote]<strong>Even more, small companies that are under control of "predators" still produce software for the Mac market... because there is money to be made doing so. Avid, for example, despite attempting to leave the Mac market, still makes hardware and software products for it. Avid very much shot itself in the foot for this, and is not in very good shape for it. Rather a lesson for anyone who would want to buy out A|W and shut down a very popular platform for its premiere (heh! pun intended) product...</strong><hr></blockquote>
I accept your Avid argument totally, and accept that it might be a lesson for a potential A|W predator.
However, if a predator's aim was to slowly "disincentivise" the Macintosh userbase, a great way to do it would be to disconnect parallel development of the WinXXX and Mac versions of a product. Then, slowly accelerate the WinXXX development so that it was superior in terms of functionality and performance, and eventually the customer makes the decision for you.
[ 06-22-2002: Message edited by: Mark- Card Carrying FanaticRealist ]</p>
"would an emulator running on a 64-bit PPC be performance competitive with real 32-bit X86 hardware?"
Answer:
No, not by a long shot.
I also see comparisons of WebObjects to .NET, which really shows a lack of understanding about what .NET is. I don't pretend to understand it fully, but it is a lot more than the WebObjects are and it will have longer term and deeper meaning than any project Apple has. My guess is that Apple would do well to provide a MacOSX based .NET implementation.
I've got to hand it to you though, you're very verbose. You probably tend to win arguments by just out-writing the opposition, don't you?
<strong>I can't believe I read the whole thing (actually I didn't, I just scanned it). One question popped out at me though:
"would an emulator running on a 64-bit PPC be performance competitive with real 32-bit X86 hardware?"
Answer:
No, not by a long shot.
I also see comparisons of WebObjects to .NET, which really shows a lack of understanding about what .NET is. I don't pretend to understand it fully, but it is a lot more than the WebObjects are and it will have longer term and deeper meaning than any project Apple has. My guess is that Apple would do well to provide a MacOSX based .NET implementation.
I've got to hand it to you though, you're very verbose. You probably tend to win arguments by just out-writing the opposition, don't you?
On the first issue, that's why I asked it as a question. But can I just ask you to expand on why the gulf would be as large as you say. Surely Virtual PC isn't so bad on a well specced G4, or is it ???
On the second issue, I wasn't even vaguely comparing where Webobjects is now to where .Net is (BTW, I do understand what .NET does), but inferring that Apple needs to widen WebObjects to become a complete alphabet soup-compatible Web Services/EAI solution so that it has a competitive offering in that market.
The options here are: -
Don't evolve WO (and watch it get eroded) or
Evolve WO and/or
Get into bed with Sun's ONE initiative as a strategic priority.
The problem with application services is that a lot of people talk about them and few of them "get" the whole picture. My personal feeling is MS have been very clever in building a quite valid image of itself as a promoter of XML as a tool for "glueing" diverse applications together, and even more clever by presenting an image that it's the only game in town.
An alternate viewpoint would be that the strength of their offering is that MS has a de facto dominant position in so many marketplaces that it is able to deliver a complete portfolio of SOAP/UDDI/etc. enabled backoffice applications (SQL Server, Exchange Server, Content Management Server, IIS, etc. ad infinitum) that can be integrated using Visual Studio .NET either with VB, or C-sharp or whatever other languages are in there.
None of the other competitors in this market have the same breadth of products to deliver; Sun can deliver the iPlanet range for directory services. collaboration, and various others but has no database; Oracle has a database and a wide variety of horizontal apps, but no directory services; SAP and Siebel have suites of horizontal apps, with modules for specific vertical markets, but no directory services or own-brand database; Apple has very little other than a neat bit of middleware that can integrate with some iPlanet products and some public-domain elements like Apache.
So the reality is that .NET is a brand image which enacapsulates MS' XML/SOAP/WSDL/UDDI efforts, but it's the breadth of products behind it that give it substance and the importance that you quite rightly identify.
However, my position is that AAPL needs to either a) find some bedfellows it can trust to deliver (I am sharply reminded of the commercial fiasco that was OpenDoc [anyone reading from Novell, thanks for all your hard work there], which I personally thought was kind of neat) if it wants to counter .NET as an MS USP or b) it needs to invent like crazy, which again you rightly identify as a potential problem given the difference in size and resources.
On another subject, I don't win arguments by outwriting the opposition. But I do try to resolve them by discussing as many points-of-view as possible. I find that normally works better than just being dismissive of someone else's position. It's called debate.
[ 06-22-2002: Message edited by: Mark- Card Carrying FanaticRealist ]
[ 06-22-2002: Message edited by: Mark- Card Carrying FanaticRealist ]</p>
On the first issue, that's why I asked it as a question. But can I just ask you to expand on why the gulf would be as large as you say. Surely Virtual PC isn't so bad on a well specced G4, or is it ???
<hr></blockquote></strong>
Processor emulation requires that each instruction in the original (emulated) program be matched with a series of native instructions which do the equivalent operations to an emulated processor state. This is because the original program was written assuming that it was modifing the state of the original processor, so the emulator has to reproduce that state and as it processes emulated instructions the actions of those are reflected on the "fake" processor state. Now add the overhead of getting the native (real) machine to decode the emulated instructions, as well as dealing with basic difference between the machines (byte-ordering being a major one). In the original 68040 emulator that Apple wrote when first switching to the PowerPC, it took about 10 PowerPC cycles for every 68040 cycle. This meant that a 150 MHz PowerPC was about the same speed as a 15 MHz 68040. Emulating an x86 won't be too far off this mark -- in some ways there is less machine state, but the instruction set is more convoluted and these days there are MMX and SSE units to deal with. The x86 is also much more memory intensive (due to the lack of registers), which doesn't really sit well with the PowerPC's load/store architecture. Also, the FPU on the x86 processors uses a floating point type that the PowerPC doesn't support (80-bits). As a result a 1 GHz PowerPC would approximate a 100 MHz x86.
One way to reduce this huge drop in performance is to identify sections of code that get executed repeatedly and convert them en-mass to a single series of "equivalent" PowerPC instructions. This works, but carries its own overhead of keeping track of when this kind of mass conversion should be done, whether it has been done, etc. Also, the emulator still needs to track the emulated machine's state. Apple's "dynamic recompliation emulator" for the 68040 reduced the performance from from about 10:1 to about 4:1. Thus a 1 GHz PowerPC would be equivalent to a 250 MHz x86. The current PentiumIV is almost ten times that fast.
Some of the performance issues can be mitigated by replacing key pieces of code with native PowerPC code that does the same thing. The biggest target is Windows or its drivers. SoftPC/VirtualPC implement some of the important pieces as real PowerPC code and thus get a real speed boost for software which relies heavily on Windows or drivers for its computationally intensive parts. Applications which bring their own computations to the party, however, are toast.
There are a bunch of other reasons why software written for one processor won't run terribly fast when emulated on another -- and software which is carefully tuned to eek out as much performance from the original processor as possible will suffer the most. That software also won't be able to take advantage of PowerPC techniques to maximize performance, so it'll suffer coming and going.
There are a couple of companies out there claiming that they have faster on-the-fly translation, but the demos have been pretty weak and their claims aren't really all that much better than this when you get right down to the nitty gritty. A 1.5 GHz can emulate a 1 GHz PowerPC when doing non-intensive stuff... oooh, wow. That's not where it really counts, is it?
[quote]<strong>
On the second issue, I wasn't even vaguely comparing where Webobjects is now to where .Net is (BTW, I do understand what .NET does)...</strong><hr></blockquote>
Okay. If Apple is going to compete in this arena, however, they should jump on the bandwagon that has the lead. As much as I hate to say it, Microsoft's song-and-dance is far more compelling than anyone else's. If Apple bought into it and did a MacOSX implementation (integrating it with their stuff, of course) that would both keep them busy and give them a fighting chance. This battle is about developer mindshare, and if Apple can be code-compatible then that is a huge leap forward for them. Beyond huge. I'm not sure about VB, but C# is a virtual machine like Java so a PowerPC native implementation could be competitive. Microsoft might just let them come along for the ride simply because that would be the straw that broke the camel's back.
Hmmm... could I fit another cliche in here? Probably not.
<strong>
Okay. If Apple is going to compete in this arena, however, they should jump on the bandwagon that has the lead. As much as I hate to say it, Microsoft's song-and-dance is far more compelling than anyone else's. If Apple bought into it and did a MacOSX implementation (integrating it with their stuff, of course) that would both keep them busy and give them a fighting chance. This battle is about developer mindshare, and if Apple can be code-compatible then that is a huge leap forward for them. Beyond huge. I'm not sure about VB, but C# is a virtual machine like Java so a PowerPC native implementation could be competitive. Microsoft might just let them come along for the ride simply because that would be the straw that broke the camel's back.
Hmmm... could I fit another cliche in here? Probably not.
Thanks for the info!
[quote] I don't object to you making an argument, but I think that they should at least vaguely observe some logic. <hr></blockquote>
[quote] HOS, a career in spin doctoring awaits you here in the UK should Tony Blair's Richlieuesque Communications Director choose to go and get a proper job. <hr></blockquote>
[quote] You're spinning again! <hr></blockquote>
<ad hominem attack>
Heh- the more you rely on ad hominem attacks, the less impressive your argument becomes... <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
</ad hominem attack>
Anyway, back into the fray:
[quote] bonus question, I believe its Monty Python (the argument sketch), but I cannot remember who the performers were. Palin & Cleese <hr></blockquote>
We have a winner! Although it's admittedly an easy lob since one would expect an Englishman to be aware of famous English comedy... <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
[quote] An example of what I'm complaining about would be: Apple currently makes mainstream systems using commoditized components which are sold at mainstream prices, therefore Apple cannot exploit the HPC computing market because they would have to retain a mainstream business model and could not therefore build a high-end system. <hr></blockquote>
Incorrect. My exact words were:
[quote] With prices between $800 and $4k, Macs are dirt-cheap... any HPC-like machine or technology that Apple employs (like a NUMA architecture) has to be cheap. Therefore, this must exclude many HPC characteristics which are too expensive to implement. <hr></blockquote>
Remember the context- this is discussing MCCFR's assumption that the Apple of today is working on or is interested in some sort of HPC or HPC-like technologies.
I stand by this statement, which is most definitely different than claiming Apple cannot ever build a high end HPC machine. I'm very clearly stating that to sell machines between $800 and $4k, Apple needs to use technology whose costs land under these magic figures. Whether Apple chooses to play in the $20k and up computing category is a completely different argument- and my arguments against that were stated in my later posts. You have my apologies if this distinction was not sufficiently clear.
Perhaps one or two more thoughts along this line to illustrate this.
Upcoming sub-$4k HPC-like technologies which could be used in Apple's $800-4k lineup include RapidIO, HyperTransport, 64-bit commodity CPUs (G5), RAID, and the kind of NUMA multiple CPU architecture discussed ad nauseum in the Dorsal M thread. These are the kind of commodity cheap HPC things I'm looking at- a high bandwidth switched bus, multiple CPU support (>2 anyway), large address space, commodity high speed networking, fast I/O.
Expensive HPC stuff which won't fit under $4k- massively parallel systems (>128 CPUs, heh, call it >4 to be more realistic), terabyte disk arrays, pervasive fiber interconnects, multi-terabyte robotic tape archival, N+1 redundancy for all components, automatic failover, pervasive massively wide busses with a direct crossbar topology between CPUs.
Thus, Apple can sell better performing machines for <$4k in the near future by borrowing some HPC concepts, independent of whether they choose (or choose not) to sell a $20k+ machine.
Continuing further, a past thread in my arguments is that this HPC-like technology _need not come from SGI_, lessening the perceived dependence Apple would have on SGI technology.
Next topic:
[quote]Using this argument, every marketplace that AAPL competes in is saturated <hr></blockquote>
Nope- an absurd abstraction ripped from context.
Apple has a dominant market share in more than one saturated market- desktop publishing comes to mind, and nonlinear semipro video editing is another. Apple has no market share in >$4k HPC, also a saturated market.
To ignore this distinction is a grievous error, as dominant market share imparts inertia. May I consider this self-evident, or should I expand on this?
[quote] The reason that companies run Sybase or Oracle on Sun Starfires is a relatively simple combination of factors: it's the only agressively marketed 64-bit general purpose business applications server with a comprehensive ISV business applications portfolio that runs a robust "open" operating system. <hr></blockquote>
This is directly ignoring my main point, which even though was quoted, I feel is so important, that I'll repeat it- most shops run Oracle on Sun, Dell, IBM, etc. hardware because Oracle has a proven, demonstrated track record of running on these platforms! Apple does not, thus giving it a significang hurdle it needs to overcome to compete here.
Re: Apps servers
[quote] I was not referring to narrow application servers, but to web application services where AAPL has WebObjects as a well-respected product. <hr></blockquote>
I don't have any real experience with WebObjects, so I'll be terse here. WO seems to be a "middleware" package which is largely restricted to being a method for developing web-based applications, in the same vein as openACS, ColdFusion, PHP, or ASP.
<a href="http://www.mackido.com/WebObjects/01-Middleware.html"; target="_blank">http://www.mackido.com/WebObjects/01-Middleware.html</a>;
Based on my experience in the "New Economy", the current Xserve will do wonders here. It will scale through multiple Xservers, rather than needing a "big iron" Xserver.
Web apps in particular are usually network or DB constrained. Most are built in redundant configurations, so multiple devices are preferable to one big device. Again, this plays into current Xserver design.
OK- my lunch break is running short, and I have a conference call I can't get out of in about 20 minutes, so I'm going to have to cut things a bit short, even though I have a lot more to say.
[quote] I want to see AAPL leverage a scalable OS on a variety of system segments all the way from Home to Enterprise!<hr></blockquote>
Ah-hah, now the end goal is being laid bare! MCCFR's dream is to have Apple be a large Enterprise vendor (in addition to its current "creative ghetto"), a la HP or Dell.
This leads into what I was asking for in one of my earlier posts, about "Apple... by Apple" vs. "SGI... by Apple."
This argument relies on:
1) There exists this thing called brand.
2) Consumers have certain associations with brand
3) Brand has value (can be positive or negative)
4) Brand has cost- to build, to maintain, or to change
More later, got to run.
-HOS
<strong>Uh-oh, things are beginning to turn ugly...
Ah-hah, now the end goal is being laid bare! MCCFR's dream is to have Apple be a large Enterprise vendor (in addition to its current "creative ghetto"), a la HP or Dell.
This leads into what I was asking for in one of my earlier posts, about "Apple... by Apple" vs. "SGI... by Apple."
This argument relies on:
1) There exists this thing called brand.
2) Consumers have certain associations with brand
3) Brand has value (can be positive or negative)
4) Brand has cost- to build, to maintain, or to change
More later, got to run.
-HOS</strong><hr></blockquote>
Firstly, I apologise for the perceived attacks - my intention was to point out that your version of "truth" is not necessarily mine or anyone elses. Likewise, the "possibility" that I'm expounding may seem like the ramblings of an idiot to some and the very crystallisation of truth and beauty to others.
We all put a spin on facts so that they support our point of view, I was just being clear in saying that was what I felt you were doing and - by comparing you to Alistair Campbell (Blair's Eminence Grise) - I was pointing out the quality with which you were executing said spin.
Also, my dream is not to have AAPL as a Large Enterprise Vendor as you put it. My dream is to have AAPL as a definitive vendor of computing solutions, no matter what the market context (Home/SoHo/SME/Enterprise/Government). Personally, I do not feel that anyone has yet rebutted this as an aspiration.
On another note: Yes, you can run 20 xServe models to deliver a web applications farm. But in a commercial world where consolidation is a pressing requirement, maybe larger businesses would like to be able to run 10-20 virtual servers on a single larger box. If you don't believe me, look at vmware.com, or join me on a project whose aim is rationalise server count on a 3-1 or 5-1 basis.
But the things that really get me about many of the arguments that the naysayers to this thread have put forward are as follows:-
The total denial that this could even be a possibility
The abject lack of ambition for AAPL to extend itself into the corporate market
The lack of confidence that AAPL could even attempt to execute such a shift in its aspirations, let alone succeed in the attempt
I've constantly said that I doubt the possibility myself: but the difference is that, despite the doubt, I was willing to explore the opportunity. Maybe Raycer or nVidia or whatever may be some part of AAPL's future. But that does not necessarily extrapolate to it being ALL of AAPL's future.
Also, whilst <a href="http://groups.google.ca/groups?q=macworld+new+york+2002&hl=en&lr=&ie=UTF-8&[email protected]&rnum=2" target="_blank">this</a> wasn't the source of my SGI theory, it throws some more sticks on my bonfire.
I would like to add that some in the Navy have Blacklisted SGI. All of the projects that I am associated with will never again consider SGI.
My point being that not all SGI Corporate recognition is positive.
Personally I think Apple would be better off buying the SGI Engineers and leaving company alone.
Kenny must die... to live
First of all I would like to congratulate everyone in this thread for contributing to what has to be the first intelligent discussion I have read on an Apple-centric forum. To have a thread such as this not descend into ignorant and misinformed Microsoft bashing, or tears and recriminations about when Apple is going to drag its core technologies into the late twentieth century never mind the early twenty first, is refreshing enough for me to contribute to my first Apple forum thread ever.
I want to return to this assertion:
Apple plus (purchased) Sgi plus (renewed co-operation with) Avid = Total domination of film and television (post-production)
Note that I have expanded this somewhat.
Firstly, let me explain my own career path, so as to establish some credentials:
Graphic Designer (on Macs when 33mhz and 16mb of ram was to die for)
to
Broadcast Designer (using Quantel Hal,Harriet and Harry, highly specialised and proprietry systems)
to
Avid Editor (Avids running on both Macs and NT)
to
Flame/Fire Operator (Discreet compositing and editing systems on SGI Octane and Onyx)
to
Post Production facility Sys Admin (running Discreet on Sgi, Avid, Shake, Maya etc on Win2k)
With Apple's recent purchase of Nothing Real and Silicon Grail, it would seem to me that they are making a fairly concerted and frankly unwelcome move into my business. I should state that the market we cater for is somewhat higher up the scale than Final Cut Pro editing DV. Think commercials with eye-watering budgets and ridiculous deadlines. My initial reaction has been to dismiss any further use of Shake and Rayz past the versions my facility is presently running (no big deal, many post houses still use Cineon, which became the core of Rayz) due to the fact that I would not even consider having a Mac in the facility (other than as part of an Avid, and even then I'd probably go with Win2k). They simply are not fast enough for what I need them to do in the time I need it done in, and I'm sure Apple is as aware of that fact as I am.
However, Apple purchasing and leveraging the technology of SGI puts me in two minds. Firstly, it would be reassuring for me for SGI to be in something of a more secure position. SGI today is Apple in the mid-nineties - fantastic technology, dismal company. Apple today is the reverse as far as I am concerned - there is only so far you can take ridiculosly lacklustre hardware in pretty boxes.
The reverse of that, however, is nagging doubt over Apple's focus. This the company that has chased market sectors over the last decade like the drunken party girl at a fraternity party, waking up in the morning with a headache, someone in bed next to them they don't recognise and no money in their wallet. Who is to say that my industry isn't simply Apple's latest night on the tiles, to be dropped when they sober up.
To return to the core of this thread, I can see incredible synergies between SGI and Apple, lets face it, Apple needs someone to teach them how to write a proper Unix for a start, by that I mean one that runs at the speed a Unix core on a RISC chip should reach. Both systems have always been known as the artist's computer. Both deal with niche markets. It's a shoe-in, and I can see it happening.
shannyla, that was a good post, sums up my thoughts on the subject. your post is a good representation of what i was saying about perceptions in the high end market, ive heard from allot of rayz users that they wont be migrating to the mac, too bad for apple. what will they do to win market segment, buy sgi perhaps. i still contend that its a great idea, cause it will make my life easier, and who dosn't want that.
<strong>
Apple needs someone to teach them how to write a proper Unix for a start, by that I mean one that runs at the speed a Unix core on a RISC chip should reach.</strong><hr></blockquote>
<img src="graemlins/smokin.gif" border="0" alt="[Chilling]" /> Whatever
it's not a purchase of SGI, but still could be cool, if it happens...
Blizaine
[ 06-30-2002: Message edited by: Blizaine ]</p>
>Greetings all.
>Firstly, let me explain my own career path, so as to establish some credentials:
>Graphic Designer (on Macs when 33mhz and 16mb of ram was to die for)
>Broadcast Designer (using Quantel Hal,Harriet and Harry, highly specialised
and proprietry systems)
>Avid Editor (Avids running on both Macs and NT)
>Flame/Fire Operator (Discreet compositing and editing systems on SGI Octane and Onyx)
>Post Production facility Sys Admin (running Discreet on Sgi, Avid, Shake, Maya etc on Win2k)
I surmise you have no unix admin experience.
>To return to the core of this thread, I can see incredible synergies
>between SGI and Apple, lets face it, Apple needs someone to teach them
>how to write a proper Unix for a start
They didn't write most of darwin, the freebsd project and next did, and
apple already has hired a bunch of people "to write a proper Unix"
(whatever that means)
> by that I mean one that runs at the speed a Unix core on a RISC chip should reach.
[you've used other unixes on PPC and know what to expect?]
Since you don't have any Unix administration experience, I think we can
safely dismiss your opinion. Perhaps you're speaking from experience with
the speed of quartz, or something else which has nothing to do with unix.
<strong>
<img src="graemlins/smokin.gif" border="0" alt="[Chilling]" /> Whatever</strong><hr></blockquote>
Whatever, maybe, but when the platform's major developers are complaining at the World Wide Developers Conference...
<a href="http://macintouch.com/wwdc2002top.html#speed"; target="_blank">http://macintouch.com/wwdc2002top.html#speed</a>;
...that the speed (or lack of...) of MOSX is making them and their applications look bad, then I would suggest that the OS is in serious need of some fettling.
The whole point of Unix OSs is that they take up tiny amounts of resources, which is why Irix is so rapid on systems with relatively low Mhz counts. One of the problems with MOSX, from my and my industry's point of view, is that is full of computationally demanding junk. Pretty interfaces and bouncing dock icons are fine and dandy for making the sale to Johnny at CompUSA, but they don't mean shit to me. The same goes for other demanding application areas such as Medical Imaging, Industrial and Architectural Visualisation and Biotech.
Now this wouldn't matter a damn to me, but Apple decideded to come crashing into my area, in a frankly gauche and careless manner. So there are reports that Pixar have swapped to Macs, I very much doubt it except for a few high-profile machines for cross-marketing purposes. Where are the high-end OpenGL cards that Maya needs? (If you say Geforce4, you are an idiot and we both know it...) Where is the truly powerful render-farm solution? Xserve isn't it, that is the most fundamentally flawed server I have ever seen. Where is the redundancy? Where is the SCSI? I wouldn't trust my mission-critical systems to IDE drives with tiny MTBFs. Add to that the requirement to rewrite everything, and I would suggest that all that has happened is that Pixar has locked the door to their Sun machine room and put up a sign saying "there's nothing here for you, move on now"