Mac Musings

Apple's Next Direction

Daniel Knight - 2 August 1999 -

A reader got me thinking when he wrote, "I find myself more and more bothered by Apple's current direction and am curious to know what you think.

"Here's what has me agitated: shutting out Be, shutting down involvement in mkLinux, the rumored 'no G4 upgrade' bomb in the G3s, OS X not compatible with pre-G3 PCI Power Macs."

Shutting Out Be

Shutting out Be was stupid from any perspective but one: Steve Jobs'. Remember, he was the visionary behind NeXT, the OS Apple bought to form the core of OS X - instead of BeOS. Steve undoubtedly sees Be's elegant OS as a challenge to the Unix core he remains devoted to.

Ditto for mkLinux - another competitor to Mac OS NeXT. Let the hackers have fun with BSD and Linux, but Apple wants one OS. Of course, this is much better than Microsoft with its abundance of operating systems sharing the name Windows.

From a strictly business standpoint, remember that Macs make up at most 10% of the worldwide computer and OS market. Letting someone gobble up a portion of that would not be good for Apple.

Remember, too, that Apple isn't just a hardware company or an operating system company. Apple sells the whole ball of wax. For everyone buying a Mac to run Linux, BeOS, or BSD, that's one less Mac OS upgrade Apple can sell in the future.

BeOS is a very promising operating system with very few available applications. It is elegant and provides all the features a stable, efficient OS should. It might have made a strong foundation for Apple's future - or it might have broken so strongly with the past that forward migration would be uncomfortable to those with years of Mac experience.

From a user perspective, it's a real shame we can't even try BeOS on anything AppleDesigned with a G3 processor.

Dropping mkLinux

It's pretty much the same story with mkLinux, a version of Linux with official Apple support. With OS X tied into a Unix kernel and Linux working almost exactly like Unix, there are two ways of looking at this.

1. Overlap between the two projects would make it easy to leverage driver design and OS components between both operating systems.
2. Why have two different Apple-supported operating systems which are so similar at the core?

By releasing mkLinux, Apple gives it a chance to live. Had Apple kept it, it would have died in the shade of OS X.

Apple needs a tight OS focus to move forward. It can't pull a Microsoft and try to support a whole family of operating systems.

The G4 Upgrade Bomb

Apparently the latest Macs have some code in ROM which may (or will) prevent a G4 upgrade from working.

The G4 will be a very powerful processor. The basic G4 has an additional math unit, allowing it to crunch numbers faster - perhaps as much as 50% faster than the already impressive G3.

The AltiVec instructions go beyond that, providing amazing support for the kind of graphic number crunching needed in Photoshop, QuickTime, and probably 3D games. (AltiVec will be in all Motorola versions of the G4. IBM may build some G4s without AltiVec.)

But both of these are small potatoes compared with multiprocessing. The new Mac OS kernels (9 and X) provide support for multiple processors within the OS. This means programs may be able to take advantage of 2 or 4 CPU G4 machines without being rewritten or recompiled.

Unfortunately, the G3 simply isn't designed to work efficiently in a multiple processor setting. However, to this point, that really hasn't been a big issue: except for Photoshop, almost nothing takes advantages of multiprocessing (MP).

However, MP can significantly improve computing power without the need for ever faster CPUs, caches, and motherboards. An OS that fully supports MP can nearly double performance by adding a second processor. (Due to the amount of processor time involved in managing multiple processors, doubling the number of processors doesn't quite double power.)

Machines like the DayStar clones harnessed up to four 604 processors, providing incredible throughput with Photoshop. The G3 outperforms the 604e at the same clock speed, but without the multiprocessing support of the 604e. Still, with the G3 approaching 500 MHz, it probably offers comparable performance to a quad-processor 200 MHz 604e system.

Imagine harnessing the power of the G4 (essentially a G3 with even faster math and MP support) at 500-600 MHz with two or four CPUs.

• A 400 MHz G3 has a MacBench 5 CPU score of about 1350.
• A 500 MHz G3 should bench at around 1650.
• A 500 MHz G4 may well bench in the 1800 to 2000 range.
• With got MP support in OS 9 and X, a dual processor 500 MHz G4 system could score in the 3000-3500 range.
• And a quad processor might be in the 6000-7000 range!

And it may do even better than that. The G4 will support a larger L2 cache (up to 2 MB), can share that cache among multiple CPUs, and has a special bus that allows CPUs to communicate with each other at full processor speed.

The beauty of the G4 is that it should be very easy to create G4 upgrades for both daughter card systems (such as the 7500) and ZIF systems (such as the Power Mac G3).

However, the "upgrade bomb" makes that impossible. Without a patch, the latest G3 systems will refuse to work with a G4 upgrade - and this appears to be a very deliberate decision on Apple's part.

Whether Apple will provide the patch or allow anyone else to remains to be seen. But it seems that Apple's desire to sell new hardware is behind the move to make the G3 systems incompatible with G4 upgrades.

OS X and Older Power Macs

Remember this is Low End Mac. I love old Macs and want to provide the best possible resources for people using older Macs.

When Apple introduced the Power Mac, it promised every Power Mac would be able to run Copland, their code name for the next great OS. This new OS would support protected memory, preemptive multiprocessing, multiprocessor support, and more.

Copland became Rhapsody, which became Mac OS X. OS X will be available in both a Server and a Client version. Although the current OS X Server runs on some older Macs, Apple says it will not support OS X Client on anything except Macs designed around the G3.

For owners of 6100s, 7500s, and clones, this is frustrating. Apple promised us features, but it looked like they wouldn't deliver.

However, Mac OS 9 (due in October) will provide some or all of the promised features to owners of older Power Macs and most, if not all, clones. It won't be OS X. It will be the end of the trail for all pre-G3 designs, at least as far as the Mac OS is concerned.

On the other hand, both OS 9 and X support the Carbon model, so for the next few years we are likely to see software designed to run on both platforms. After that, owners of pre-G3 Power Macs and clones will join Mac Plus, PowerBook 180, Mac IIci, and Quadra owners as OS orphans.

This is frustrating, especially since OS X has a Mach 3.0 kernel at the core. This kernel, used with operating systems such as BSD and Linux, has also been written to support older hardware.

On the other hand, there may be unsupported (by Apple) hacks that will allow use of OS X on the older hardware. This would require serious hardware support work from the Mac community, but could keep older Power Macs OS-current into the 21st century.

It would be nicer to see Apple provide this support, but Apple seems less supportive of their legacy customers than they did in the past.

Closing Thoughts

Apple will not support older hardware with OS X. However, between OS 9 and possible hacks allowing use of OS X on officially unsupported hardware, these legacy Power Macs and clones may reap the features long ago promised by by Apple.

The question is, will OS 9 deliver enough of those promised features to keep Apple off the hook, or will we see a class action suit?

I think that's the primary reason Apple is developing OS 9 instead of waiting until OS X is ready to ship. They know the lawyers will have a field day of the older Power Macs don't get the promised OS features.

The future is OS X, not BeOS, Linux, or Windows.

BeOS will thrive as a niche OS. Rumors indicate it will be the OS of choice for a new line of Amiga computers. The marriage of BeOS with a video-centric computer will be wonderful.

Linux remains a geek OS. You have to know too much to install and use it, although some companies are simplifying that. Still, with OS X you can reap the benefits of Unix with a shell that works just like a Macintosh.

Windows is nearly collapsing under its own weight. I hope a lot of people burned by the Y2K fiasco will learn their lesson and never again trust a Microsoft OS. Beyond that, Microsoft has fragmented its market into Windows 95 (still the #1 OS), Windows 98, Windows CE, Windows NT 4.0, Windows 2000, Windows 98 Second Edition, and who knows what other variants down the road.

On top of that, Windows is seriously lacking in the stability department, which is a leading reason for defection to Linux.

Beyond Mac OS X, Apple needs to continue cloning the Macintosh - which is exactly what they've done with the iMac and iBook. By continuing to offer innovative designs which can be manufactured elsewhere, Apple can grow sales volume and market share without investing in new production facilities.

Current niches Apple should address:

1. Subnotebook. This could be either a handheld or thing-and-light Mac OS computer. (Not a Palm, and not a Newton.) The great expectation was that the iBook would be both small and inexpensive. Apple did okay on pricing, but needs a small portable.
2. Power Tower. With the release of G4 systems anticipated in early 2000, Apple needs to offer a powerhouse system with multiple CPUs, more bays for removable drives, and at least six expansion slots. It matters little how expensive it is or how well it sells -- Apple needs this one to convince video professionals that they are serious about supporting the power user.
3. Inexpensive. Nice as the iMac is, Apple could probably sell a lot of compact, modular Macs. Make them small like the Quadra 630, give them one or two PCI slots, and let the users choose their own monitors. Apple could probably sell a competent entry level monitor-free computer like this for $750 or so.

Apple has great future prospects, but will have difficulty achieving them if the established base of Mac users feels abandoned.

They did an incredible job moving to the PowerPC processor five years ago. They did a very competent job retaining backward compatibility with both System 7.0 and Mac OS 8.0. They embraced USB and forced the industry to support it; they're doing the same with FireWire.

If Apple can continue to offer backward compatibility with software and most peripherals, provide all the "buzzword compliant" features promised for older Power Macs in OS 9, and make OS X look and feel enough like the current Mac OS that we feel totally comfortable migrating to it, they will keep current Mac users happy.

If, at the same time, they can fill other hardware niches, they can also continue to grow market share.

Managing change is no easy. It's too easy to alienate people. But I'm convinced Apple can do it without losing the current Mac community.