graphic artists can do with practical, high-res, 3-D animation on platforms such as Silicon Graphics' workstations.
On a less obvious note, analysts both scientific and financial have benefited from the throughput and large memory capabilities of the 64-bit workstation.
But what about the average user? Are we all now going to navigate
through 3-D models just to answer our e-mail? I hope not. My real desk is messy enough; I don't need to sift through virtual clutter as well.
Multimedia would seem to be an obvious benefit of the new architecture: It was the original killer app for high-end 32-bit computing. But we're at a crossroads. Today's computers are basically good enough for local video. Tomorrow's will be dependent on sufficient bandwidth for network video, and that will arrive well after the 64-bit business desktop.
One of the best uses of this power would be to cross the last mile to true real-time speech-to-text synthesis. The best products ever are on the market right now, and they're just short of acceptable for continuous, speaker-independent use.
Large memory applications -- an unsexy but useful concept -- are another great use for the new architecture. Think of it as an obese client; you run an entire multidimensional database in RAM on your own desktop. Answers to very complex questions appear in seconds.
The new chips will have a huge impact on servers. I just spent three days at a data-warehousing summit conducted by Dr. Richard Hackathorn for Lockheed Martin (Hackathorn often writes about databases for BYTE). I heard from organizations as diverse as Citicorp, NASA, and US West, which are already running multiterabyte data warehouses. The large memory addressing of 64-bit computers, coupled with their higher throughput, will help many businesses approach those multiterabyte levels with inexpensive symmetric multiprocessing servers.
The same phenomenon applies to Web servers. Do CGI scripts, Java servlets, and database access overwhelm your Web server? Next-generation chips should be a quantum leap in price/performance.
Finally, let me make a plea for a new way to use all this power. The average business desktop has all it needs to run typical productivity apps. Let's take 25 percent of all the new cycles over today's 32-bit, 300-MHz Pentium II desktop and use them for intelligent self-manag
ement. I recently learned from Lucent that a typical large corporate PBX has 5 million lines of code and that 55 percent of it is self-administering. According to Lucent, 80 percent of software problems are resolved by the system itself. OK, a PC is more versatile than a PBX, but I'd like to see that design philosophy take hold in the near-supercomputers you'll have on your desk in the 64-bit era.
Perhaps the greatest significance of the new IA-64 architecture lies in the opportunity to do some urban renewal on the kludgey landscape of the x86. The x86 architecture started when microcomputers were still a hobbyist item. Since then, PCs have carried a lot of baggage and entailed a lot of compromises. Let's hope that Intel uses the clean slate of a new architecture to the best advantage.
New technology means new risks, though, and never before in the history of the PC has the risk been larger. Intel, along with Microsoft and many application vendors, will have to provide as much backward compatibili
ty as possible. The new Intel architecture is probably a bigger departure from the past than Apple's switch from the 68000 to the PowerPC. After all, the PowerPC used RISC technology -- a known quantity -- while Intel will use a totally new architecture. We'll certainly keep you informed as the details become clearer.
Mark Schlack, Editor in Chief,
mschlack@bix.com
Step Aside, x86
