bradandkathy.com

Power Mac Musings, Part 3

The Great Unknown

Note: This review originally appeared in the January 1995 issue of the Southern Maine Apple Users Group (SMAUG) Newsletter. It has been modified slightly from the original version, but no effort has been made to update information that has become out-of-date since the original publication.

This is the last in a series of articles about the Power Macintosh, RISC technology, and the future of the Macintosh. Two months ago, I discussed the advantages of RISC microprocessors over CISC designs and how they achieve such fantastic performance. Last month, I filled you in on some of my Power Mac experiences over the last nine months or so. This month, I’ll round things out with a discussion of what lies ahead for the PowerPC and Power Macintosh.

As I mentioned in the first article, the PowerPC 601 processor in the current crop of Power Macs is just the first in a family of similar RISC processors. At the time the PowerPC alliance was announced, there were to be four chips in the PowerPC family, starting with the 601 and moving on (and up) to the 603, 604, and 620. The engineers at the joint Somerset design facility in Texas had set out a challenging timeline of delivering all four chips by the end of 1995. Thus far, they’ve delivered on their promises. The 601, 603, and 604 are currently in production at various levels and the 620 has recently reached “first silicon,” a chip fabricators’ term meaning it has moved beyond pure design to be fabricated for the first time. If you’re wondering what happened to the 602, you’re not alone! If you find out, let me know.

The 601 is considered by some to be a “transition” processor from IBM’s earlier POWER design to the PowerPC family. As such, its blazing speed only hints at what lies ahead. The 604, expected to show up in Macs sometime in mid-1995, is the first pure PowerPC chip. Depending on the application, it could boost performance as much as 100% over similar 601s. The 620, which is a full 64-bit microprocessor, could nearly double performance again! It will show up in servers and high-performance workstations sometime in 1996.

The 603 is a different story. Its overriding design goal was to provide a low-power PowerPC design for notebook computers and other low-power applications. Its performance is similar to the 601, but it achieves that speed with considerably lower power consumption. Apple is working on creating the next generation of PowerBooks based on the 603+, a slight variant with a larger cache and a few other tweaks to improve performance. Although Apple has already shown prototypes of the PowerPC-based PowerBooks (PowerPowerBooks?), they are not expected to hit the shelves until sometime next summer. I had a chance to see one of the early prototypes in action several months ago and I must say that seeing a tiny PowerBook run circles around behemoth desktop machines was quite a sight.

All this talk of its siblings may make you think you can overlook the lowly 601, but that’s far from true. Last month, Apple introduced the Power Macintosh 8100/110, a 110 MHz version of the 8100 that delivers 20-30% better performance than the 8100/80. The 110 MHz 601, the first “601+” chip to hit the streets, uses new chip- fabrication technology to shrink the size of the chip, reducing power requirements, and allowing faster clock speeds. IBM has already demonstrated 601+ chips running as high as 120 MHz, and further speed increases may well be possible. As faster chips become more available (and cheaper!), I expect them to show up throughout the Power Macintosh lineup. The 6100/60 is likely to move up to 66 MHz (the current speed of the 7100), bumping the 7100 up to 80 MHz, all with little or no increase in price.

But processor speed is not the only change coming to the Power Macintosh. Apple is planning changes in both hardware and software that will improve the speed and usability of new Macs. Future Power Macs will abandon the NuBus expansion architecture in use since the introduction of the Mac II back in 1988 for the greener pastures of PCI (Peripheral Connection Interface). PCI is an Intel standard bus already in use in high-end 486 and Pentium PCs. Besides offering the potential for true cross-platform expansion cards (that is, cards that can be used in both PCs and Macs), PCI supports much higher speeds than NuBus. As desktop video and other high-bandwidth applications enter the mainstream, this speed will be greatly appreciated. The PowerPC partners recently announced another upcoming hardware improvement, the Common Hardware Reference Platform (CHRP) specification. This specification, which will be public once it is finalized in early 1995, will outline a standard PowerPC-based computer architecture. Apple, Motorola, and IBM plan to build future machines to the CHRP spec. Any CHRP-compliant PC will be able to run several operating systems, including AIX (IBM’s version of UNIX), OS/2, Windows NT, and most important to you and me, MacOS.

The potential importance of this specification cannot be understated. Provided nothing goes horribly awry, sometime in 1996, you will be able to buy compatible PowerPC systems from Apple, IBM, Motorola, or a plethora of others and run your choice of operating system (MacOS, of course!). For the first time in their history, Mac hardware and Mac software will no longer be stuck together with superglue. Just as users of today’s Intel clones can choose from a great variety of hardware manufacturers and several operating systems (Windows, OS/2, Linux, NextStep, etc.), PowerPC users will be able to do the same. This is a drastic step for Apple, but one that will likely be good for both Apple and its customers.

On the software side, Apple’s improvements will come in several stages. According to the latest rumors I’ve heard, sometime in the first half of 1995 Apple will release the next major revision of its system software. Code-named “Marconi,” this release will incorporate several major improvements including support for PCI and the PowerPC 603 and 604 chips. In addition, it is rumored to include a much-improved version of the 680x0 emulator for Power Macs and a great deal more native PowerPC code. Less relevant to this discussion (but interesting nonetheless), Marconi will also include the first system-level support for OpenDoc, Apple’s new document-centered approach to computing and will include a revamped communications architecture known as OpenTransport.

Following Marconi sometime in late 1995 will be a much more significant improvement in the Mac’s operating system (MacOS). This release, code named “Copland,” will support a new (to the Mac) method of dividing processor time among multiple running applications. Among other improvements, this method, known as preemptive multi-tasking, will keep any one application from hogging all of the processor’s time or freezing the whole machine. Combined with another improvement, protected memory, this should greatly reduce the number and severity of crashes and other instabilities. Under Copland, an application crash that might bring the whole Mac screeching to a halt today would have little impact on other running applications (or the system as a whole). Copland will also make better use of internal multi-tasking (known as threading) within the Finder, so that, for example, you could copy a file, format a disk, and launch an application all at the same time. Some of these improvements rely on hardware features of the Power Macs (and AV Quadras) to help do their magic.

Copland is likely to be nearly 100% native code on the Power Macintosh. More native code means more speed, so even today’s Power Macs should see a speed boost with Copland installed. In addition, Copland will add oodles of other features and improvements that aren’t specifically relevant to the Power Mac alone.

Finally, sometime in 1995 or 1996, Apple will introduce the “Gershwin” operating system. Details are especially sketchy of Gershwin, but it will probably include support for multi-processor machines, sophisticated 3-D graphics, and software “agents” that can step in to perform common tasks for you. Most important, however, will be Gershwin’s ability to run on different platforms, such as the latest from Intel, or Digital. This represents the last step in freeing the Mac operating system from one specific hardware platform.

Well, that’s about it for the world of PowerPC! I hope you’ve enjoyed reading this series of articles as much as I’ve enjoyed writing them. If you have any questions or comments, feel free to contact me by e-mail at bdm@sprylo.com. For now, I’m just sitting back waiting for my new Power Macintosh 9550/150 with four PowerPC 620s running the Gershwin OS. I’ll let you know when it comes in.

[Go to Part 1] [Go to Part 2]