Wednesday, June 08, 2005

Apple: To boldy go to Intel while leaving PowerPC

Blogger's Note:


New life for the old 'Star Trek' project
By CNET News.com Staffhttp://news.com.com/New+life+for+the+old+Star+Trek+project/2100-1045_3-5734189.html Story last modified Mon Jun 06 14:41:00 PDT 2005


The possibility of running the Mac operating system on Intel PCs isn't new. In his book "Apple Confidential 2.0," author Owen Linzmayer recounted a drive in the early 1990s to do the same thing--a project dubbed "Star Trek." (Indeed, the idea was initially broached as far back as 1985). The chapter, titled "The Star Trek saga," is reprinted here with permission from Linzmayer and publisher No Starch Press.

The idea of porting the Mac OS to run on Intel processors wasn't new (Dan Eilers, Apple's director of strategic investment, first proposed the idea in 1985), but it gained a renewed sense of urgency after Apple shipped System 7 in May 1991 and it failed to make headway against Microsoft Windows 3.0. Ironically, it wasn't a determined Apple engineer or insightful executive who finally got the ball rolling. That honor goes to a company that few outside of the industry have ever heard of: Novell of Provo, Utah.

Networking giant Novell wanted to provide an alternative to Windows by creating a Mac-like interface for its DR-DOS that ran on Intel-based IBM PC clones but feared getting sued by Apple (at the time, Apple's copyright infringement suit against Microsoft was still very much alive). Rather than risk an infestation of lawyers at its headquarters, Novell decided to find out whether Apple was willing to work together on such a project. On Valentine's Day 1992, Darrell Miller, Novell's VP of strategic marketing, met with several Apple software managers to reveal his company's plan. Excited by the possibilities, the Apple contingent obtained CEO John Sculley's blessing, and the two companies immediately began working together on a project that came to be called Star Trek, because it would boldly go where no Mac had gone before: the Intel platform. When Bill Gates heard of Apple's plan to put the Mac OS on an Intel machine, he responded by saying it would be "like putting lipstick on a chicken."

A group of 4 engineers from Novell and 14 from Apple was put together by Gifford Calenda in suite 400 of a Novell marketing office called Regency One in Santa Clara, directly across the street from Intel's headquarters. That was no accident. Sculley had met with Intel CEO Andy Grove, who agreed to help the Star Trek project because, ever paranoid, he didn't want to be so dependent on Microsoft. Each Star Trek engineer was given an office with a Mac and a 486 PC clone donated by Intel.

On July 17, the Trekkie team was given until Halloween to come up with a working "proof of concept." As an indication of the importance Apple placed upon the project, each engineer would receive a bonus of between $16,000 and $25,000 if they succeeded. "We worked like dogs. It was some of the most fun I've had working," recalls team member Fred Monroe. Free of managerial meddling, the small team not only succeeded in getting the Mac's Finder to run on the PC clones, it also managed to get QuickTime and some of QuickDraw GX working, as well as the "Welcome to Macintosh" startup greeting. Having met their deadline, the Trekkies collected their bonuses and took off for a well-deserved vacation in Cancun, Mexico. In their minds, they had laid the groundwork for a product that could save Apple by allowing it to compete head-to-head with the inferior Microsoft Windows on its own turf: Intel-based computers.

Now it was up to team leader Chris DeRossi and Roger Heinen, VP of software engineering, to convince Apple's executive staff that Star Trek was worth pursuing. On December 4, they presented the Star Trek prototype to the assembled staff, many of whom couldn't believe their eyes. From all outward appearances, here was the fabled Mac OS running on an Intel computer; Star Trek had managed to penetrate deep behind enemy lines. Fred Forsyth, head of Apple's manufacturing business and hardware engineering, saw his career flash before his eyes. If Apple was successful in getting the Mac OS to run on Intel, demand for Apple's hardware would likely slump. Furthermore, the company was committed to moving the Macintosh to the PowerPC, and the Star Trek project was perceived to be a threat to that effort as well. How would it look to partners IBM and Motorola if Apple was porting the Mac OS to Intel processors at the same time it was collaborating on the PowerPC? Over these objections, Heinen was given the go-ahead to have his team attack the detail work to make Star Trek fully functional.

Armed with the executive staff 's approval, Mark Gonzales, the project marketing manager, made the rounds of PC clone vendors to gauge their interest in bundling Star Trek on their systems. Most were intrigued, but argued that they couldn't afford to pay much for it because their contracts for Windows 3.1 forced them to pay a royalty to Microsoft for every computer shipped, regardless of what operating system it contained. (This anticompetitive practice eventually landed Microsoft in trouble with the Department of Justice.)

Worse than the clone makers' tepid reception to Star Trek was Heinen's defection to Microsoft at the beginning of 1993. Without Heinen around to protect the Trekkies, in February the project was moved back onto Apple's campus at Bandley 5 and placed under the control of David C. Nagel, then head of the Advanced Technology Group. The project ballooned from 18 people to 50, and most were forced to write detailed specifications and white papers instead of concentrating on writing code. Then COO Michael Spindler instituted a round of belt tightening. After Nagel had allocated his budget to revising the OS for the PowerPC and updating System 7, there was not enough money left over to fund the completion of Star Trek, which was estimated would take 18 months and $20 million by some accounts. Nagel considered merging Star Trek with another project just getting under way. Code-named Raptor, it was an alternative to the Pink OS mired in the bickering at Taligent and was intended to run on any CPU, not just Intel or Motorola processors. However, the merger plan was deemed infeasible, and Star Trek disappeared into a black hole in June 1993. Work on Raptor would continue and eventually evolve into Copland, the long-promised, never-shipped sequel to Mac OS (see "The Copland Crisis").

Even if the engineers had managed to complete Star Trek, it wasn't as if suddenly every existing Mac application could run on Intel computers. Star Trek was designed to be source-level compatible, not binary compatible, with the Mac OS, meaning Mac applications would have to be recompiled by their developers to run on Intel processors. Those programs that directly addressed the Mac hardware would have to be rewritten. Needless to say, many software publishers were skeptical about the amount of work that would be necessary to port their products to PCs running Star Trek. Besides, a working demo of Star Trek isn't the same thing as a finished product. Remember, Apple engineers cobbled together pretty impressive proof-of-concept demos for Copland and Pink, too, but they never shipped either.

Just because Apple docked Star Trek doesn't mean you can't run Mac applications on other computers. In October 1995, Apple said that adapting the Mac OS for IBM's PowerPC Reference Platform (known as PReP) would be a technically daunting challenge and used that as an excuse to promote its variation, CHRP. That didn't stop Quix Computer, a company outside Lucerne, Switzerland, from succeeding in making the Mac OS work on PReP systems with just a half-dozen engineers. Despite the potential widespread demand for such a product, Apple refused to license the Mac OS to Quix, and as a result, Quix moved on to develop technologies for more receptive companies. Apple couldn't quash Executor as easily. A small company called ARDI in Albuquerque, New Mexico, managed to reverse-engineer the Mac OS and Toolbox to create a version of System 7.0 called Executor that can run on Intel 486, Pentium, and DEC's Alpha processors. Executor is far from a perfect port. It doesn't support PowerPC code, it has no serial port access (meaning AppleTalk and modems won't work), sound input/output is primitive, there are no provisions for internationalization, and neither extensions nor control panels can be used. Nonetheless, Executor runs many standard Mac programs on PCs. Imagine what it could have done with Apple's support.

In late 1997, many of the Star Trek engineers held a reunion in Cupertino following the expiration of their five-year non-disclosure agreements. Rumor had it that there was a new Star Trek NG (Next Generation) project involving an emulator that allowed Mac OS applications to run on Intel hardware under Rhapsody. As the years have passed with no new word on Star Trek NG, it appears to have gotten lost in space again.

Copyright ©1995-2005 CNET Networks, Inc. All rights reserved


Apple fails to meet (a columnist's) expectations
By Michael Kanellos http://news.com.com/Apple+fails+to+meet+a+columnists+expectations/2010-1071_3-5734187.html Story last modified Mon Jun 06 14:43:00 PDT 2005

Who ever thought Apple would convert to Intel?

Not me. Eighteen months ago, I wrote that such a deal wouldn't occur because it would potentially allow clone makers to produce cheap Macs and make price comparisons easier between Macs and Windows-based PCs. (I also predicted that IBM services would begin to teeter, so I wasn't a complete bonehead.)

When news of deals began to surface last month, I wrote another article quoting several analysts downplaying any budding relationship. They say animals can sense things early, but I completely ignored the fact that two weeks ago my cat started drinking coffee and fiddling with the band saw.

Still, the predictions weren't that bad. It won't be an easy transition. Sales will likely decline between now and June 2006 when the first Intel-based Macs come out, because few will want to buy a computer out of a product line heading for the tar pits. Clever PC makers and software developers, particularly in emerging markets, will be able to figure out how to put the Mac OS on white-box PCs.

Nonetheless, many decisions are fraught with risk and difficulty. The Austro-Hungarian empire originally didn't enforce its will on the Serbs in 1914 because of the all-consuming importance of the sheep-cheese trade in the Balkans. But fear of losing prestige, worries about Russia and pressure from the Kaiser in Germany all pushed them toward a precarious course of action.

With PowerPC chips, Apple found itself grappling with an ever-widening performance gap. Although PowerPC was somewhat comparable to Intel/Advanced Micro Devices processors in the first part of the decade, those chips began to pull away. Intel and AMD also came out with new, faster, cheaper versions at a faster clip.

The troubles began to come to a head with the G5, the IBM processor that debuted in 2003. Benchmarks from Apple touting the superiority of the G5 were widely questioned.

Apple also began to include a liquid cooling system in its PCs for the processor. Apple fans claimed the cooling system underscored the company's visionary sense of engineering. Dan Hutcheson, CEO of VLSI Research, saw it differently. The liquid system added about $50 to manufacturing costs. Historically, liquid cooling is also a sign that the processor line won't be around for long.

When Apple began to publicly complain about the G5 on conference calls, the companies had entered a danger zone. IBM doesn't cotton well to criticism, particularly from allies. "We invented the cash register! We came up with fractals! We sponsor Christmas specials," is sort of IBM's attitude toward the rest of the world. Big Blue started to hang out with game console makers and listen to old Springsteen records on the car stereo. Divorce suddenly loomed as a possibility.

"As we looked out toward the future, we envisioned the kind of products we'd like to build, but we didn't know how to build (them) with the future of the Power PC road map," Apple CEO Steve Jobs said at the company's developer conference.
Oddly enough, Apple has been gaining market share in the past few quarters despite the G5 issues.

Intel has power consumption problems with the Pentium 4 too, but a shift to dual core cures some of that. Future notebook and desktop chips (Yonah, Merom, Conroe) will further reduce power consumption, which in turn will let Apple come out with sleek, small systems.

There will be other benefits too. Intel increasingly is performing more of the internal engineering work required in getting servers and other computers out the door. It makes reference designs and lines up component manufacturers to adopt certain specs. By taking advantage of some of this work, Apple can likely reduce the delta between its computers--which tends to be about $300--and a similarly configured computer from a PC company.

On the other hand, Apple will have to live with being one of many. The red carpet was unfurled Monday, but a year from now, Apple will be lumped into the same courtesy shuttle as Toshiba, Sony, Gateway and the other "not in the top five" PC makers. Acer sells more computers than Apple and therefore will likely qualify for higher volume discounts.

Apple will also lose one more aspect of its uniqueness, which the company seems to crave, so who knows what will happen next? Feeling a bit dejected, Apple may start casting about again. Look, there's Hector Ruiz of AMD, the company might say to itself. He talks quite a bit about the importance of the emerging market. They're sort of an underdog....

Copyright ©1995-2005 CNET Networks, Inc. All rights reserved.


Apple throws the switch, aligns with Intel
By Ina Friedhttp://news.com.com/Apple+throws+the+switch%2C+aligns+with+Intel/2100-7341_3-5733756.html Story last modified Mon Jun 06 11:04:00 PDT 2005

SAN FRANCISCO--After years of trying to get people to switch to Macs from Intel-based computers, Apple Computer itself has switched.

CEO Steve Jobs announced Monday that Apple will gradually shift its Mac line to Intel-based chips over the next two years. The move confirms a timetable first reported by CNET News.com.

Jobs' announcement formed the centerpiece of a keynote speech to Mac programmers attending the company's annual Worldwide Developer Conference here. The conference, expected to draw some 3,800 attendees this year, is a traditional venue for Apple product launches.

In his speech, Jobs revealed that Apple has been developing all versions of OS X since its inception to run on Intel and PowerPC chips.

"Mac OS X has been leading a secret double life the past five years," he said.

The move to Intel marks a tectonic shift for Apple, which has used processors from IBM and Motorola (now Freescale Semiconductor) throughout the life of the Mac. However, the company has changed architectures before, shifting in the 1990s from Motorola's 68000 family of chips to the PowerPC architecture jointly developed by IBM and Motorola.

Jobs also noted the significant effort required earlier this decade when Apple moved from Mac OS 9 to Mac OS X. Although the operating systems are only a digit apart, he noted that the move to a Unix-based system was a major shift. "This was a brain transplant," Jobs said.
The CEO showed a demo of the Tiger operating system on an Intel-based machine, saying, "We've been running on an Intel system all morning."

As for why Apple was making the shift, Jobs pointed both to past problems and to the PowerPC road map, which he said won't deliver enough performance at the low-power usages needed for powerful notebooks.

Two years ago at the same conference, Jobs introduced the first G5-based Power Macs and promised developers that the company would have a 3GHz PowerMac within 12 months. The company still doesn't have a machine that fast. "We haven't been able to deliver," he said. Nor has Apple been able to introduce a G5-based laptop--something Jobs said "I think a lot of you would like."

Things weren't looking better in the coming months, Jobs said, saying that IBM's PowerPC road map would only deliver about a fifth the performace per watt as a comparable Intel chip.

Jobs said there are a lot of products Apple envisions for the coming years, but "we don't know how to build them with the future PowerPC road map."

Jobs added that most of the necessary OS work has been done, but developers will have to do some work to make their applications work on Intel-based machines.

Transcoding tool to the rescue
Programs written will require various amounts of effort--from a few days of tweaking to months of rewriting--depending on the tools used to create them.

Some software that's insulated from the underlying chips, such as widgets and Java applications, will work without modification, Jobs said.

Going forward, Mac developers will be able to create universal binaries of their programs that will run on both types of chips.

In the meantime, Apple has a transcoding tool called Rosetta that will allow programs written for PowerPC chips to run on Intel-based machines. "Every application is not going to be universal from Day 1," Jobs told the audience.

A Microsoft executive said the company would create universal binaries with future versions of Office for the Mac. And Adobe Systems CEO Bruce Chizen told developers they can be "absolutely sure" his company would support Apple's transition.

"The only question I have, Steve, is: What took you so long?" Chizen said.

Also on Monday, Jobs said the next version of OS X, called Leopard, will be released in late 2006 or early 2007. That is the same time frame as Microsoft's next Windows update, dubbed Longhorn, he noted. Microsoft has said Longhorn will be released by late 2006. After Jobs' presentation, Apple Senior Vice President Phil Schiller addressed the issue of running Windows on Macs, saying there are no plans to sell or support Windows on an Intel-based Mac. "That doesn't preclude someone from running it on a Mac. They probably will," he said. "We won't do anything to preclude that."

However, Schiller said the company does not plan to let people run Mac OS X on other computer makers' hardware. "We will not allow running Mac OS X on anything other than an Apple Mac," he said.

Copyright ©1995-2005 CNET Networks, Inc. All rights reserved.

The brains behind Apple's Rosetta: Transitive
By Stephen Shankland
http://news.com.com/The+brains+behind+Apples+Rosetta+Transitive/2100-1016_3-5736190.html Story last modified Wed Jun 08 04:00:00 PDT 2005

A Silicon Valley start-up called Transitive is supplying Apple Computer with a crucial bridge to enable the move to Intel-based computers, but skeptics worry about performance problems that have plagued similar products.

Transitive is providing the engine used in Apple's Rosetta software, which translates software for its current machines using PowerPC processors so it can run on forthcoming Intel-based Macintoshes. "We've had a long-term relationship with them," Transitive Chief Executive Bob Wiederhold said Tuesday.

As a program runs, Rosetta translates its PowerPC instructions into corresponding x86 instructions. Although there are limits to what programs it can translate, the software promises to ease the transition that current Apple customers and software developers face. Apple Chief Executive Steve Jobs on Monday demonstrated Rosetta during a keynote address, showing it running PowerPC versions of Adobe Photoshop and Microsoft Word and Excel--three applications essential to the success of the Macintosh line.

News.contextWhat's new:A Silicon Valley start-up called Transitive is supplying Apple Computer with a crucial bridge to enable the move to Intel-based computers.

Bottom line:Apple and Transitive face performance challenges. Success has been elusive for computer makers trying to support one chip's software on a machine with a different chip.

Jobs' Rosetta demonstration went smoothly--he loaded and edited several documents--but both Apple and Transitive face performance challenges with Rosetta. Success has been elusive for computer makers trying to support one chip's software on a machine with a different chip.

"History says that binary translation basically doesn't work," said Illuminata analyst Gordon Haff. "The day may come when someone can do a good enough job with it, but that concept has been thrown out there many times in the computer industry, and it's always fallen flat on its face."

But Los Gatos, Calif.-based Transitive is willing to set high expectations when comparing software compiled natively for the new processor to that compiled for the older processor and running on the new one.

In the case of Transitive's first customer, Silicon Graphics Inc., software for the older processor generally reaches at least 80 percent of the speed of native software, Wiederhold said. But that high score stems partly from the fact that the SGI systems are used for graphics tasks, which have little or no translation penalty, he said.

With more computationally intense tasks, the performance of translated software is between 60 percent and 80 percent of native software, Wiederhold said.

Another skeptic is Nathan Brookwood of Insight 64. "Everybody always has said 50 (percent) or 60 percent and delivered 30 (percent) or 40 percent," he said. Among those who have tried: Digital Equipment Corp.'s FX!32 to run x86 Windows programs on computers with Alpha chips; Hewlett-Packard's Aries software to run HP-UX software for PA- RISC chips on Itanium; and Intel's IA32-EL software to run software for x86 chips on Itanium.

Jobs was satisfied, though. During his demonstration, Jobs said translated software runs "pretty fast," though his presentation's slide said performance is "good (enough)." His demonstration computer had a 3.6GHz Pentium 4 and 2GB of memory.

Apple, though not known for bending over backward to support users of older Macs, has some experience helping users with processor transitions. When it changed from Motorola 680x0 processors to PowerPC in 1994, it included emulation software that would let users run the older software on the newer machines.

And Wiederhold is delighted to have Apple as a customer. "Like many start-up companies with breakthrough technology, there's a lot of skepticism about the technology itself--whether we can meet the claims we discuss," he said. "Getting proof points out there is very important to our success."

One thing that's unclear is whether Rosetta will work in the other direction--translating x86 software for use on PowerPC Macs, something that could significantly expand Transitive's revenue sources. That feature, by ensuring future Mac software will work on older-generation machines, could help convince potential PowerPC-based Mac customers not to put off their purchases.

Transitive last fall released a version of QuickTransit that would support such a feature, but Wiederhold wouldn't comment on whether Apple plans to use it.

However, Apple hopes programmers will create what it calls "universal binaries"--software that includes versions for both processors in one package. Doing so would significantly increase the size of a program, but if programmers followed this practice, an Intel-to-PowerPC translator wouldn't be needed.

Apple has been reluctant to discuss where Transitive fits into the Rosetta technology, though Jobs did confirm in a New York Times interview that Transitive is playing a role.

In an interview with CNET News.com, Apple Senior Vice President Phil Schiller declined to say how much of Rosetta was developed in-house. "I'm not going to talk about details, but it's Apple technology," Schiller said.

Transitive has about 65 employees, with all engineering staff in Manchester, England. Founder and Chief Technology Officer Alasdair Rawsthorne developed the technology in 1995 at the University of Manchester and built a company around it in 2000.

The company has raised $24 million in three rounds of investment--in October 2000, in February 2002 and in September 2004, Wiederhold said. Investors are Pond Venture Partners, Crescendo Ventures and Accel Partners.

Six major computer makers are Transitive customers, and some new ones should be announced in coming months, Wiederhold said. Later this year or early next, the company wants to start selling products to a second class of customer: software companies that can include QuickTransit as a quick way to bring products to new processors.

QuickTransit can be used to bring software to Itanium, PowerPC, and x86 from Intel and Advanced Micro Devices. It can translate software from computers with x86, mainframe, Power or MIPS chips.

AltiVec and other limitations
While Rosetta will work to translate many Mac programs, it does have some important limitations.

"Many, but not all, applications can run translated," Apple said in a paper for developers.

"Applications that run translated will never run as fast as they run as a native binary because the translation process itself incurs a processing cost."

Apple said that Rosetta is "designed to translate currently shipping applications that run on a PowerPC with a G3 processor and that are built for Mac OS X."
Complete coverageApple's new core

However, Apple said Rosetta can't run several types of code: that written specifically to use the PowerPC's AltiVec instructions; that which requires a G4 or G5 chip; programs written for Mac OS 9, which today can run with Mac OS X's "classic" environment; kernel extensions; applications that depend on kernel extensions; and code that inserts preferences in the System Preferences pane.

"How compatible your application is with Rosetta depends on the type of application it is," Apple said. "Applications that have a lot of user interaction and low computational needs, such as a word processor, are quite compatible. Those that have a moderate amount of user interaction and some high computational needs or that use OpenGL are, in most cases, also quite compatible. Those that have intense computing needs aren't compatible."

Apple said that there are no visual indications that an application is being translated. A dialog box in the Finder can be used, though, to see whether an application exists only in a PowerPC binary.

Copyright ©1995-2005 CNET Networks, Inc. All rights reserved

0 Comments:

Post a Comment

<< Home