Jon Udell
INSIDE THE POWERPC REVOLUTION, Jeff Duntemann and Ron Pronk, Coriolis Group, ISBN 1-883577-04-7, $24.95
The PowerPC family of processors has rocketed to prominence atop two powerful boosters. One is technological. These speedy processors exemplify practical RISC design. The other is cultural. The PowerPC has become a lightning rod for those weary of Intel and Microsoft's near-monopolistic control of desktop computing. Like the revolution it chronicles, Inside the PowerPC Revolution mixes technology and rebellion.
On the technological front, there's plenty of engaging and informative discussion of RISC versus CISC, pipelining, branch prediction, multiple execution units, primary and secondary cac
he architectures, chip fabrication, and how these subjects relate to the PowerPC 601 and 603. (There's some coverage of the 604, which was announced as the book went to press.) All this will be useful to readers looking for a clear, accessible primer on RISC.
The authors point out, for example, that there's nothing minimalistic about RISC. The PowerPC, as is typical of the RISC genre, has plenty of instructions, some rather complex. There's nothing ``reduced'' about RISC. The PowerPC merely exemplifies what the authors call ``microprocessor new art''--abundant registers, simplified addressing, fixed-length instructions, specialized execution units, and pipelining. A solid analysis of Intel's Pentium shows why it's vulnerable, not only to RISC competitors like the PowerPC, but also to alternative 80x86 offerings like Cyrix's M1.
On the cultural front, things sometimes get heavy-handed. Take the inexplicable claim that the 80x86 instruction set is ``unfriendly to object-oriented operating systems.
'' Say what? The IBM/Apple/Motorola alliance has a lot going for it, but to anoint microkernel-based PowerPC systems that IBM and Apple have yet to deliver as the ideal platforms for object computing is just plain weird.
A fascinating survey of IBM's previous RISC efforts--the early 801 project, the ill-fated RT, and the RS/6000--reveals lessons learned along the way and puts the PowerPC in a historical context. But the authors don't place the PowerPC in a modern context. There's no discussion of Mips, Alpha, SPARC, PA, or other RISC alternatives.
On Apple's side of the PowerPC coin, that's probably OK. After all, Power Macs did vault Apple, almost overnight, to the top of the heap of RISC-system vendors. The authors tell this incredible Cinderella story with relish, describing Apple's new machines in detail and not flinching from revealing the software compromises that got the product to market so quickly.
On the flip side of the coin, though, there's an awful lot of hand waving in suppo
rt of IBM. With no corroborating evidence, the authors all but proclaim victory for PReP (PowerPC Reference Platform) systems running IBM's still-embryonic Workplace operating system and human-centered user interface, while deriding ``Bill Gates' Windows NT drivel.'' This unnecessary antagonism detracts from an excellent discussion of the operating-system options for PowerPC systems and the strategies those operating systems will use to emulate legacy Mac and Windows software. On balance, though, you'll learn a lot about the history, politics, technologies, and commercial prospects of PowerPC-based systems, and about RISC in general.
Jon Udell is a BYTE senior technical editor at large. You can contact him on the Internet or BIX at
judell@bix.com
.
PowerPC: Cultural and Technological Perspective
