Guest Viewpoint: To PPC or not to PPC

As the Amiga prepares to move into the future, questions about processors for the next generation, and how to implement them, continue to be raised

By Tim Molinder, Contributing Writer, tmolinder@earthlink.net

Now that the question of who will own the Amiga has been resolved attention must turn to the future of the platform. All of the Amiga's components, after many years of neglect, need upgrading. The processor, graphics, sound, and OS all combine to determine what price and capabilities users expect. This is the first in a series of articles that will explore each of these systems and speculate on directions Amiga International, AI for short, and third parties might take based on my own thoughts and opinions as a professional Amiga user since 1988. The CPU more than any single factor determines system speed. The Motorola 680x0 series powers the Amiga but since the '060 is the last of this series a new chip must step up to replace it. The major contender, the PowerPC, was anointed successor by Amiga Technologies before the Escom bankruptcy and remains the front runner. With Phase 5 already well on the way to delivering the first PPC-powered Amiga accelerator cards, what path should the Amiga take to its new processor home?

Follow the leader. Apple has made a successful transition through software emulation. This allows a clean break with the 680x0 and gives developers a strong incentive to port their software. The emulator allows current applications to run and all that is required is designing a new A4000 processor card and new A1200 motherboard.

A second approach has been proposed by phase 5 with its PowerUp boards. A 680x0 chip and PPC chip are combined on the same board allowing current code to run at native speeds while new software can access the PPC. Phase 5 has done a significant amount of work in providing a PowerPC development environment and ported versions of its CyberGraphX software. A PPC-native Workbench and PPC-native application software, while the logical next step, are not required for users to benefit from PPC acceleration. This allows more time to make significant upgrades to Workbench and AmigaDOS while developers can quickly bring PPC-accelerated applications to the market.

Both approaches have advantages. A clean non-680x0 approach through software emulation sounds attractive but is fraught with peril. Apple got lucky and numerous developers ported their code to PPC. The emulator barely worked and even when it did programs ran slow. Are Amiga users going to buy a new $2000 Power Amiga if it runs Lightwave or Imagine slower than a stock A1200? While Apple had a healthy market and aggressive developers to push the users through the transition quickly the Amiga has an anemic market and relatively few active developers. How many will spend the time or money to port to PPC until the number of users becomes a viable market?

In contrast the phase 5 dual processor PPC boards have the advantage of running existing applications as fast or faster than previously while providing a path to even greater speed with PPC-native code. Developers can transition parts of programs which can benefit the most from the PPC's speed right away without having to port the whole code. More importantly development can begin immediately without a complete a PPC port of Workbench and AmigaDOS. While phase 5 has developed libraries to allow software to use the PPC this is not a completed port of the Amiga operating system. Even more exciting are the possibilities of PPC and 680x0 processors working together with the PPC spared from the task of running the OS and free to devote all of its speed to applications code. In fact this may be a stepping stone to a multi-processor Amiga. Is there a downside? A problem arises if developers continue to produce 680x0-only code and the incentive to port Workbench and AmigaDOS wanes. The alternative is waiting for the PPC versions. But is it important to have a PPC Workbench and AmigaDOS since Workbench runs at a speedy clip even on an '020? It's applications software like Lightwave and ImageFX that need the speed boost. Each successive version of the OS would become more and more PPC-native until eventually all of it runs on the PPC. The vacancy left by the departing 680x0 processor could be filled with a second PPC making the Amiga a dual-processor system like the BeBox. A radical application of this approach would be the addition of other types of processors. If a 680x0 and PPC can coexist why not a DEC Alpha or Pentium chip or even systems with all three. Windows 95, Windows NT, Linux, and MacOS could all run together as tasks of the Amiga Operating System much as Emplant and ShapeShifter run MacOS. Users could use any application regardless of the operating system. The PowerUp board proves that the core OS and applications need not use the same processor or type and software tasks can be divided between them. Not only is this the best way for the Amiga to transition to the PPC but also points the way to a new future for the Amiga.

As a footnote an upgrade letter arrived from Impulse this weekend as I was finishing this article. It described the features in the upcoming Imagine 6.0 and one little line caught my attention: PowerPC support! Developers are catching PPC fever and while no specific mention was made of phase 5's boards, since they are the only announced PPC devices for the Amiga, it seems logical to assume that the reference is to them. It's an exciting thing to see a developer who was on the verge of canceling its Amiga version come roaring back into the market. While Imagine has its flaws I encourage all past and present users to give it a serious look and upgrade as I have. Kudos to Impulse!