Conclusion: Picking Winners and Losers
It is tempting to see an active period of fierce competition and change in the computing industry as a prelude to a Darwinian shake out of computer architectures and vendors from the marketplace. But history has always shown that such battles are almost always inconclusive. Although the writing is clearly on the wall for the MIPS architecture in the workstation and server market, it appears to have a bright future ahead of it in embedded control applications in telecom and consumer markets. MIPS wasn’t defeated in the competitive high end 64-bit MPU marketplace due to technological problems, or the superiority of other processor families. MIPS’s defeat occurred due to a self-fulfilling prophecy initiated by a crisis of confidence in the minds of SGI executive management, who were unwilling to invest in new processor core development out of the fear they couldn’t keep up with IA-64. Score one for Intel’s trademark ability to engage in, and win, campaigns of psychological warfare against competitors (which is rivaled perhaps only by Microsoft).
IA-64 processors will arrive on the 64-bit MPU scene and gradually grab market share. But this will be due more to who makes them rather than how they perform. And this share will largely come from IA-64 filling the voids left by MIPS and (eventually) PA-RISC and helping to grow the 64-bit market in the direction of the standard high volume server and workstation hardware niche currently occupied by Intel’s 32-bit Xeon product line. Like the claims made for the Intel i860 a decade earlier, the alleged superiority of IA-64 and EPIC will prove to be illusionary except for a tiny subset of computing problems. And there is little reason to believe that existing RISC processor families such as Alpha, POWER, and SPARC will not continue to be competitive and thrive in a rapidly growing 64-bit MPU marketplace on the basis of their current strengths.
Discuss (15 comments)