Powers of Two

A few weeks ago, an ENT cover story revealed that Microsoft had dropped support for NT on Compaq's Alpha 64-bit processors. I bit my lip a little as I read about it. Something about the indifferent tone of the report bothered me, but it wasn't until I brought it up later that day in a conversation with a colleague that I started to think about it.

"What's the big deal, anyway," he asked, after I launched into my familiar tirade about Digital Equipment Corp.'s -- which developed Alpha and has since been bought by Compaq -- legendary incompetence at selling the advantages of its 64-bit architecture. "After all, except for a few scientists, who really needs 64-bit computing," he said.

I stopped for a minute and did the math in my head. My friend instinctively assumed that 64-bit computing was twice as good as 32-bit computing. In a digital world accustomed to orders of magnitude technical improvements every year, doubling hardly seemed worth a second thought. Too long out of high school, my friend had forgotten that two to the 64th power isn't just two times two to the 32nd power, its 2 to the 32nd power squared. The reality is that 64-bit computing is potentially four billion times more powerful than 32-bit computing -- a crucial fact that today's network services designers need to consider in their future plans. The historic evolution of the coming 64-bit computing revolution explains why.

Back in the '70s, eight-bit microprocessors ushered in the age of the PC, offering memory space of a few kilobytes. Back then, ASCII text and numbers were king, and the day's microcomputers proved up to the task of processing reasonably large text messages in the form of simple word processing documents and modest spreadsheets. As the market for these toys grew, lots of ordinary people began using microcomputers to communicate with each other. Eventually, people grew bored with just text and numbers.

In response, by the '80s, PC software had developed a limited capability of handling more advanced media, such as recorded sound. Contemporary 16-bit processors, offering basic addressing for 64K of RAM, with mapping extensions of up to a megabyte, provided enough room to store and manipulate a few minutes of audio information. This whetted the world's appetite for multimedia communications.

By the '90s, people began producing movies with their PCs. Early crude video processing capabilities blossomed with the availability of 32-bit processors, providing up to 4 GB of RAM space in which to play. Let's take a moment to put this into perspective. Whereas a 16-bit processor could store and process a few minutes of recorded speeches, a 32-bit processor -- equal to the square of a 16-bit computer's capabilities -- could represent a complete, hour-long, full-motion audio and video rendering of a television biography.

Which is why, as we enter the next century, 64-bit computing architecture promises so much for the world we'll soon all be living in. Equal to the square of today's 32-bit processors, tomorrow's 64-bit machines will, theoretically, be able to store an hour-long documentary for almost every human being on earth. As corporate network designers are tasked with the job of taking their companies' products and services directly to a worldwide market, 64-bit processor addressing will become as important a consideration as RAID array size and processor speed in sizing a network server.

We used to joke that Digital Equipment Corp.(DEC), if they sold sushi, would market it as "cold, wet, dead fish." With any luck, Intel and Microsoft will do a good job developing a reliable 64-bit system platform that we'll all soon be needing -- and do a better job than DEC did explaining why we need it. --Al Cini is a senior consultant with Computer Methods Corp. (Marlton, N.J.) specializing in systems and network integration. Contact him at al.cini@computermethods.com.