Mac OS X and the Cell Processor: A Marriage Made in Heaven?

In discussing the real world potential of the new Cell processor (jointly developed by Sony, IBM, and Toshiba), Robert X. Cringely points out that software, not hardware, will determine the fate of any new computing platform.

Software, not hardware.

Repeat that mantra to yourself. It goes a long way in explaining the ongoing survival of Apple's Macintosh computers and their nearly complete failure to make inroads into a world dominated by Microsoft Windows.

Ground Zero

Apple, Radio Shack, and Commodore started the personal computing revolution on an equal footing with the introduction of home computers. Prior to 1977, microcomputers were for hobbyists who cobbled together their own computers; from 1997 forward, the Apple II, Commodore PET, and Radio Shack TRS-80 created a market for ready-to-use personal computers.

There were no operating systems. There was nearly no software. Everyone was on a level playing field, and each computer came with a version of BASIC that allowed buyers to create their own programs.

As the personal computer market grew, two thing happened. The big one was the introduction of affordable floppy drives, which created a much broader market for commercial software. PC owners could buy a word processor and just use it - and that also meant that their next PC would probably be the same brand so it could run the same programs and use the same files.

With one exception, personal computers of the 1977-1980 period used proprietary operating systems, floppy disk formats, and hardware. Apple, Atari, Commodore, Sinclair, Tandy, and others made the whole widget.

The exception was CP/M (created in 1974 by Gary Kildall), an operating system that worked on the Zilog Z-80 and Intel 8080 and 8085 CPUs. Any computer that used CP/M could use the same software and the same files - but there were a multitude of incompatible floppy disk formats that often made it a challenge to move files from one computer to another.

Microsoft made a Z-80 card so the Apple II could run CP/M, and the operating system was also ported to the TRS-80.

Enter IBM

In 1981, IBM entered the personal computer market, and things have never been the same. The original IBM PC didn't come with a floppy drive, a video card, or even an operating system - they were all extra-cost items.

A typical IBM PC system included two 5.25" 160 KB single-sided floppy drives, a video card, one or two I/O cards (for printers, modems, etc.), and PC-DOS 1.0. The operating system, also known as MS-DOS, pretty much clones CP/M for the new IBM architecture, and IBM's choice of an Intel 8088 CPU made it fairly easy to port CP/M software to the new platform.

The IBM name, the ability to leverage the installed base of CP/M software, and the availability of MS-DOS on non-IBM hardware helped make this the standard platform for the computing industry. Companies that made DOS-compatible hardware incompatible with IBM's floppies, memory architecture, or expansion bus may have made a better product in some respects (the TI Professional Computer and Zenith Z-100 come to mind), but being unable to read IBM disks or run unmodified software designed for the IBM doomed them.

IBM's architecture and Microsoft's operating system created a platform that decimated the competition in the business world and with computer hobbyists. CP/M was considered a relic of the past by 1984.

Enter the Macintosh

Apple unveiled the only alternative hardware platform to survive against the IBM/Microsoft platform in January 1984. The Macintosh used a 3.5" floppy disk, which was incompatible with DOS computers and Apple's other computer lines. It was based on a 68000 CPU, which was different from the 808x, 6502, and 6809 CPUs in the other computers on the market.

In short, it was different. Very different. Completely different. And it was an uphill battle to gain a foothold in the Microsoft-dominated market. Apple worked with Microsoft to make sure Word and Multiplan (a predecessor to Excel) were available on the Mac - even today it's crucial to the Mac's survival that Microsoft Word and Excel run on Apple hardware.

Still, by 1984 there was a wealth of software for the IBM and its multitude of clones. Getting people to switch was unlikely, and the Mac has never had anything close to the number of software titles of the Microsoft DOS/Windows platform.

It was niches that grew the Mac's market, and none more than desktop publishing, a field invented by Aldus PageMaker and the LaserWriter printer. Because of PageMaker, LaserWriters, and Apple's very simply networking, Macs became the platform of choice for page design. This later extended into other fields, such as illustration (Freehand, Illustrator) and imaging (especially Photoshop).

The Mac has been successful in the music world as well, offering good MIDI support early on. It was also the first computer line to offer a CD-ROM drive.

And it was always the second-place operating system as IBM/Microsoft built in the CP/M base and then built Windows on the DOS base, always leveraging yesterday's standards to keep tomorrow's customer.

Apple Survival

There are enough applications for the Macintosh that most people can do everything they need to do on the platform. That's been true for nearly 20 years now. And Apple has done a marvelous job leveraging their own history to grow and maintain the Mac market.

Thanks to a very clever 680x0 emulator, Apple was able to move to the PowerPC CPU while allowing pre-PPC software to run on the new hardware. In fact, many parts of the operating system remained in 680x0 code until after the Mac OS stopped supporting 680x0 hardware.

The Macintosh owes its survival to Macintosh users who followed the platform through hardware and software changes, including the transitions to PowerPC and to OS X. Apple has been very successful selling 3 million or more computers a year, losing only a few users to the Windows platform and probably gaining as many from Windows as it lost.

The Future

Cringely is looking at the future of the Cell processor, which is linked to the PowerPC architecture that has powered Macs since 1994. It's more different from today's Mac CPUs than the G5 is from the G4, and it remains to be seen how useful (or useless) it might be at the heart of a computing platform.

If it has the flexibility and power - and it certainly appears that it will - it could form the basis of a new generation of personal computers that could go head-to-head with the Intel/Windows platform.

Microsoft has tried to leverage Windows' dominance in the past, porting Windows NT to PowerPC and a few other CPUs, but they discovered the same obstacles there that others had in standing against the Microsoft standard - existing software wouldn't run on the new Windows hardware. That made migration difficult and killed Windows on other hardware platforms.

Today there are three computing platforms: Intel/Windows, PowerPC/Mac OS, and Unix/Linux on almost any hardware out there. How are they positioned for Cell-based computing?

Windows on Cell

Microsoft would face exactly the same problems with Cell architecture that it did when porting Windows to other hardware platforms. The simple fact that existing versions of Word, Excel, PowerPoint, and the like won't run on non-Intel architecture means that every program would have to be redone for the new hardware.

I can't see Microsoft going that route, because they know few would follow them to the new platform.

*nix on Cell

Linux, Unix, and BSD have survived in part because they are hardware independent operating systems. You can run Linux on a PC, on a Mac, or even on an iPod. The key is that *nix and *nix applications have to be compiled for each platform. You can't just take a program that's compiled to run under Linux on a PC and have it run on a Mac; you have to recompile or rebuild the software first.

For *nix geeks, that's not an issue, and you can rest assured that Linux will be available for Cell architecture soon after the first Cell-based computers ship.

While programs to replace Word, Excel, Photoshop, Quark, and the like exist for *nix, the whole *nix paradigm assumes users who really know what they're doing - that was one of the biggest arguments against OS X ever succeeding. Geeks may adopt Linux on Cell, but the masses won't until Linux is as easy to set up and use as Windows or the Mac OS.

OS X on Cell

The most promising candidate to control a generation of Cell-based hardware is Mac OS X. It's already based on PowerPC architecture, and the obstacles Apple has dealt with in moving from Motorola 680x0 CPUs to early PowerPC CPUs and more recently to the G5 put them in a good position to adapt to a somewhat different architecture.

Further, Mac software is already designed to run on PowerPC hardware, so I suspect a lot of it could be coaxed to run without modification on the new Cell architecture. For best performance, software will have to be updated to take advantage of the new hardware, but thing should run well enough to make it feasible to run current OS X programs on the new computers.

OS X for All

The next step is to create an OS X consortium so Apple could work with Sony, IBM, and others to create a standard Cell computing platform that others could license. There are already PC companies interested in licensing the Mac OS, whether because of Windows problems or Microsoft's hubris, and a new hardware platform would give them the opportunity to "go Macintosh."

Apple would have to be willing to give up complete hardware control over the platform, which is why a consortium would be essential to the success of OS X on Cell. And Apple would probably see a decline in Macintosh hardware sales due to competition, but that would allow Apple to concentrate on what it does best - innovation and easy-to-use products.

The Mac OS already has almost all the software anyone could ever need. The Cell architecture would address one area where the Mac is weak - 3D gaming - and could help reshape Apple as the Macintosh/Mac OS/iTunes/iPod/great and easy applications company.

Just imagine 10 million new Mac OS users a year. Millions more OS X upgrades. Millions more people who might buy Pages, iLife, Final Cut, and Apple's other software. Millions more who might buy Apple hardware the next time around, since their Cell-based HP or Compaq already runs the same OS and programs.

After all, it's the software that sells the hardware, not the other way around.