Slashdot Mirror
Apple, Scully, And Intel vs. Motorola
fsharp writes "I've heard too many comments suggesting that Apple should have moved to Intel (x86). The Register has an article exploring John Scully's recent comments about his failure to move the Mac to x86. Scully critiques his decisions based entirely on hindsight, and in doing so, identifies Dell as a the chief competitor and the way Apple could have slayed this evil dragon would have been to move to Intel early on in his tenure. Not so fast. Hindsight can be 20/20, however it can also be quite myopic if one suffers from selective memory. The article does a good job of examining the options available at the time when Apple rewrote the MacOS for the Intel x86. How safe a bet or great a risk would it have been for Apple to switch, given the quality of chips offered (at that time) from Intel?"
Its killed KDE/LINUX!. Its fun to trick people into thinking I've got a macintosh!
All the Apps are there
Itunes = JuK
Safari = Konqueror
Finder = Konqueror
Dock = Kicker
Menubar = Kicker
bbedit = Kate
Quicktime = Kmplayer!
Appleworks = Koffice!
Had nothing to do with the processor architecture they chose to go with; that was nearly invisible to the end user, and the only things switching to x86 could possibly have helped with was possible binary compatibility with dos/windows, and better a better perception of performance in the mid to late 90s. Apple's key error (in hindsight) was the failure to commoditize their hardware at the right time, probably in the mid '80s, when apple had a reasonably useful graphical OS far more advanced than dos, an "OS" which could've been put together in a week by a competent hacker. Commoditization would've led to multiple manufacturers competing (and enlarging the market, with their advertising dollars) on costs and hardware features, rather than a lone company trying to maintain multiple product lines and losing focus in the process. Apple could've emerged as a really viable alternative to Microsoft, and could in fact be in the same position MS is in today (but with fewer security holes along the way, and a saner platform ;-). But I prefer apple the way it is now, fairly lean and mean and focused. I think there's a non-zero probability that the apple culture would've led to complete self-destruction beyond a certain company size. Sculley did an okay job considering his background, but he's a hypocrite (as is SJ) who has no real clue about technology, although he had a better grasp than Gil Amelio (read his autobiography, his "tech" comments are hilarious).
"The slave who knows his master's will and does not get ready...will be be beaten with many blows."Luke 12:47-48
At the time that the Mac moved to PPC there is no way that Intel could have offered such a seamless transition. The ability of the PPC to emulate the 68k at good speed was what made the transition work. If Apple had tried to move to the x86 platform they would have lost a lot of customers with a big investment in software as the x86 could not emulate the 68k at usable speed.
Using an alternative archetecture has also allowed Apple to hold on to its uniqueness, which has in turn guarded it somewhat from fierce comparison to the x86 crowd.
No-one in the PC business saw Dell coming, if Apple had been just another x86 vendor with a nice OS they would be facing the same problems as HP, Compaq etc did when confronted by Dell's better supply chain model.
I think that Sculley is being remarkably revisionist in his views. The article points out a lot of the folly in his musings.
The G5 and the relationship with IBM is more than enough to now justify the choie of the PPC architecture.
I want to use these Mod points but I can't find anything Interesting, Informative or Insightful on Slashdot.
No. You're only discussing about a 15 year old frame, here : if Microsoft had switched to 680x0 or to the PPC 10 to 15 years ago (or any other reason that would have made the motorola platform a more dominant one), then it would have been improved in the same way than the x86 did but it would also slow down after gaining a monopole.
Now, Apple is backed by IBM and they already got G4, then G5 to run on multi-processor consummer computers.
I guess, we're only in the beginning of this story and I consider the variety is what will keep our market evolving.
Trolling using another account since 2005.
In 5 years Apple will be primarily a music sales company anyway, with their mildly successful niche computer/OS sales department just kind of chugging along as it always has in the background..
I'm not going to dump $1500 or so for the priviledge of getting some overpriced, proprietary hardware platform.
... and all the reasons for running Linux (open source tools) are just as valid now under OSX, so ...
I did. And I do not regret it. I've been a computer-user since the 70's, and as far as I'm concerned this tiBook has been the best computer I've ever invested in, and I've owned many. I've gotten more done with this than I ever did with a PC, I've had *NO* virus problems, no crashes (really, not a single system-failure type crash in the 3 years I've had it), and it has been around the world with me, twice, and still keeps on running.
I had to work on PC hardware today for a few hours. Man, I'm so sick of having to deal with PC hardware problems.
Just give me a machine that works, a decent operating system, and get out of the way. Thank you, Apple.
That said, if I could find an Intel laptop with the same design as a tiBook, I'd probably run Linux or FreeBSD on it - but, frankly, I doubt it would be more cost effective to do so than just upgrading to a newer 17" alBook
; -- the corruption of government starts with its secrets. a truly free people keep no secrets. --
Selected quotes from a confidential memorandum from Bill Gates to John Scully of Apple dated June 25, 1985.
Source: Wired Magazine, November 1997, page 126-128.
A memo on "Apple Licensing of Mac Technology."
Apple's stated position in personal computer is innovative technology leader. This position implies that Apple must create a standard on new, advanced technology. They must establish a "revolutionary" architecture, which necessarily implies new development incompatible with existing architectures.
Apple must make Macintosh a standard. But no personal computer company, not even IBM, can create a standard without independent support. Even though Apple realized this, they have not been able to gain the independent support required to be perceived as a standard.
The significant investment (especially independent support) in a "standard personal computer" results in an incredible momentum for its architecture. Specifically, the IBM PC architecture continues to receive huge investment and gains additional momentum [...] The investment in the IBM architecture includes development of differentiated compatibles, software, and peripherals; user and sales channel education; and most importantly, attitudes and perceptions that are not easily changed.
Any deficiencies in the IBM architecture are quickly eliminated by independent support [...] The closed architecture prevents similar independent investment in the Macintosh. The IBM architecture, when compared to the Macintosh, probably has more than 100 times the engineering resources applied to it when investment of compatible manufacturers is included. The ratio becomes even greater when the manufacturers of expansion cards are included.
Conclusion:
As the independent investment in a "standard" architecture grows, so does the momentum for that architecture. The industry has reached the point where it is now impossible for Apple to create a standard out of their innovative technology without support from, and the resulting credibility of, other personal computer manufacturers. Thus APPLE MUST OPEN THE MACINTOSH ARCHITECTURE TO HAVE THE INDEPENDENT SUPPORT REQUIRED TO GAIN MOMENTUM AND ESTABLISH A STANDARD. [emphasis mine]
The Mac has not become a standard:
The Macintosh has failed to attain the critical mass necessary for the technology to be considered a long term contender.
[...]
Recommendation:
Apple should license Macintosh technology to 3-5 significant manufacturers for the development of "Mac Compatibles".
US manufacturers and contacts: ideal companies - in addition to credibility, they have large account sales force that can establish the Mac architecture in larger companies:
- AT&T, James Edwards - Wang, An Wang - Digital Equipment Corporation, Ken Olsen - Texas Instruments, Jerry Junkins - Hewlett Packard, John Young
Other companies:
[ list of companies and contact names deleted ]
Apple should license the Macintosh technology to US and European companies in a way that allows them to go to other companies for manufacturing. Sony, Kyocera [...] are good candidates for OEM manufacturing of Mac compatibles.
MICROSOFT IS VERY WILLING TO HELP APPLE IMPLEMENT THIS STRATEGY. We are familiar with the key manufacturers, their strategies and strengths. We also have a great deal of experience in OEMing system software.
Rationale:
1. The companies that license Mac technology would add credibility to the Macintosh architecture.
2. These companies would broaden the available product offerings through their "Mac-compatible" product lines:
- They would each innovate and add features to the basic systems [...]
-
You know, you can buy a supercomputer that's even faster than a G5, and even more expensive too!
Interesting that you mention this, because I am currently examining that. I have a rack of 6 dual-P3s that range from 800mhz to 1ghz. I have structured that rack into a cluster that does genetic algorithm / genetic programming processing in Java.
I am a Linux guy, generally, but like the polish of "OS X" on the desktop, so I bought one of the new G5s as a development platform. Last week, I did a test run on my G5 and discovered something I did not expect; The dual G5 ran generations faster than my cluster!!!
This is no-bullshit, for real. I am not making this up. The rack was taking 20-30 seconds per generation (with a total of 6 machines and 10 processors, each running 800mhz-1ghz). The G5 was processing generations in 17seconds. Granted, the P3s in the rack are not the newest intel Xenons or anything and the G5 does not have to go through the overhead of distributing work and collecting results, nevertheless, the results took me by surprise.
I am running on Panther (7B85), if that makes a difference to anyone (it is much faster than 10.2.8). After seeing this and scratching my head about it, I certainly came to understand VT's decision much better.
This is directly relevant to your comment, BTW. Certainly a rack of Xenons would tear the 17second number to peices, but each Xenon is the same or higher price than a G5, so we can't really do that comparison.
The rack of lower-price intel commodity hardware just doesn't cut it.
The company has to decide: is SUN a hardware company? that would mean investing a lot in the development of SPARC, killing the Solaris x86 line and fighting Linux
Well, you can buy a 1U Sun SPARC now for less than a comparable 1U Dell. Sun is price-competitive now at the low end, for the first time. And, they have an edge Dell can't match, namely that Sun guarantees forwards compatibility - you can take an old app from old hardware and an old Solaris, and run it on the latest kit and OS. If you can't, and you stuck to documented APIs in your code, Sun treats it as a bug in Solaris and fixes it. Dell can't provide that because they don't control any of the OSs they offer. Not only that, but Sun's hardware is of a higher quality. I recently experienced my first Sun hardware failure in years - two SPARC processors failed in a production server. The machine just kept right on running, as it was designed to do, and we swapped a new board in when it was convenient. What can you hot-swap on a Dell these days, just disks?
What I'm getting at is that Sun is not a hardware company nor a software company. It's a platform company, like IBM's mainframe division, and like Apple. If you want to compare it to Dell, a more subtle analysis is required than just "price per box".
---- The Open Source Record Label : : LOCARECORDS.COM
You're forgetting the third way Apple could do itself a favour. They could open source Aqua. Imagine the possibilities: 1. I'm pretty sure that a project would have Aqua running on Linux or FreeBSD in a few months. In the meantime, Apple gets free development and bugfixes for Aqua, and when it's finished, we get the modern, excellent X Windows replacement (with a backwards compatible rootless X Server) people have been wanting for years and years. 2. Apple can still sell their own hardware, and they can limit key products like iTunes to PPC binary. This way, they don't loose out to comoditised x86 boxes. 3. Microsoft won't be pissed and stop producing Mac Office (which is important to a lot of people on the Mac), because again they can just produce PPC binaries instead of x86 binaries. Instead we get OpenOffice.org for Aqua (hopefully faster after they dump the horrendous toolkit they're using now).
I agree, it is impressive.
I wonder if anyone's done a serious test to see, for example, what is the best number crunching performance you can get for $6000. I guess the intricacies of setting up a distributed system would be rather prohibitive. It would also be cool to see how different types of application would go - even though overall the memory would be slower on the P3s there would be a heck of a lot of cache if you add it all together, which counts for something.
Blades would be nice... plus your geek-cool would be through the roof with those things. Nice to see IBM still making serious computers IMHO, evil super-corporation or not.
Read Pynchon.
in late 2000 when Motorola was beginning to show how they were going to fuck up G4 production in the future
Let me channel text from Apple Recon from that period:
Jobs had a meeting at Motorola. Topic of discussion was the 'working relationship'.
Jobs had cancelled the Mac clones. This meant that Motorola has lost 87 million dollars directly in the shutdown of the clone line. In addition to screwing up the %age of PPC chips that would be produced and put into general purpose computers.
Instead of the over 15%, Motorola was left with 5% of the chip volume going to Apple. Motorola execs were not happy. Jobs response? 'It will be great in 2 years when we won't be using you!' (Remember, at this time there was: Blue Box - Mac OS 6-7-8 API Yellow Box - The NeXTSTEP API Red Box - NeXTSTEP still on Intel) Sometime after this meeting, Red Box was 'quietly' dropped. And Motorola - they ended up introducing 'altavec'(sp) op codes. Ever look at those Op codes? They are more DSP centric than general purpose computing centric. And Apple 'hyped' them as a 'feature' - like killing Red Box the hyping was to attempt to smooth over things with Motorola.
Meanwhile, the comment 'in 2 years we are dropping you' stayed with Motorola, and they positioned the chip to work better in the remaining 95% of the market. Why should have Motorola spent time in improving the chip for a customer who was going to leave?
Thus Jobs's hot headded comment to Motorola cost Apple PPC preformance and Apple had to kill the promised public continuation of the NeXTSTEP product on X86.
Most of the 'Apple zealots' forget Jobs's meeting with Motorola.
I have one quibble with this - is the x86 market still expanding? In the developing world, I suspect pretty much everyone who needs a computer already has one, and unless they do video or games their raw hardware is fast enough for what they do (word processing, spreadsheets, email, web browsing).
If this is true, then what we're left with is competition of style (ease of use, fit and finish) versus externalities (compatibility with the rest of the universe). If there's any justice, Apple ought to continue to grow.
Though I hope they don't "win" - a monopoly Apple would probably become just as fat, lazy, and obnoxious as Microsoft.
To a Lisp hacker, XML is S-expressions in drag.
not to mention that the G5 now beats the penitiums in every area: raw cpu perfromance, vector processing, wide data busses, and hyper transport. Plus its loping along at a mere 2Ghz and a rather small chip area while the x86 technology is sweating bullets (and heat) at 4Ghz trying to keep its pipelines full and branch predictions correct and its massive chip real-estate in sync on the clock signal.
The itanium technology of very long instruction compiling is falling on its predictive race condition petards with two speed rollbacks to date.
The PPC software world has yet to truly exploit those sexy well thoughtout vector ops (altivec) and that insanely fast bus, so there's lots of legs for improvement even without processor speed bumps in the near future. the small chip are will end itslef to muliple uints per die, the future still looks bright.
meanwhile the only real developments in the x86 world is the transition to 64 bit (and maybe when the motherboards start to catch up, to ubiquitous hypertransport). But perhaps its nearing the end of its speed, chip area, coherent vector instruction set life cycle?
Some drink at the fountain of knowledge. Others just gargle.