Today, devotees of computer-aided design, digital photography, and 3-D games want all the speed they can get. Mainstream PC users have relatively little use for the performance boost, but within a year or so, they'll be buying Pentium III or its descendants anyway, as Intel gradually but steadily moves this CPU design from the top to the bottom of the PC market. Throughout the history of the PC, except for the past two years, most applications severely taxed our systems, creating a seemingly insatiable demand for faster PCs. Recently, however, the masses have been sated. Systems based on Intel's low-cost Celeron processor, AMD's K6-2, or Cyrix's M II are fast enough for most users. This is not to say that using computers is not rife with waiting and wasted time, but these drains come from elements beyond the processor. For many users, the thing that slows them down is figuring out how to use the computer and what to do at each step of the way. Once they give a command, rarely is there much of a delay for it to execute. When there are long delays on the computer end, they are not due predominantly to the CPU. Accessing the Internet is a prime example, and faster processors aren't going to help here. We simply need higher-speed network connections. And other tasks that seem slow, such as opening or closing a program or a document, are generally limited by the drive transfer rate. One of the slowest functions on a typical PC today is simply booting up. Intel and Microsoft have been working for years to reduce the bootup delay to a few seconds by using a suspend mode,rather than a true off state. The fact that it has taken years for the PC industry to implement this technology, even though it has been feasible for a long time, shows just how much inertia the industry has when it comes to anything other than improving megahertz and megabytes. Disk speed is gradually improving, but the biggest reductions in time spent waiting for the disk will come from better use of main memory (DRAM). As mainstream PCs move into the range of 128MB to 256MB of memory in the next couple of years, there is room to keep much more information in DRAM. Today's operating systems and applications, however, are written for a world in which main memory is tiny, and information is frequently swapped to and from the disk. More intelligent software, redesigned to take advantage of huge amounts of DRAM, could eliminate many of the drive delays that plague today's systems. So is there any need for faster processors? The classic PC applications--word processing, spreadsheets, Web browsing, e-mail, and most other office applications -- run fast enough so that speed is not at the top of users' minds. And will mainstream applications that require more processing power begin to emerge? Eventually, this will happen, but the question is when. If you compare the abilities of today's software to what you'd like your computer to do ideally, it is still primitive. Unfortunately, there doesn't seem to be any software on the verge of shipping that will drive a surge in the demand for high-end processors. During the next year or so, Pentium III will be popular primarily with three classes of PC buyers: corporate PC buyers who want to maximize the useful lives of their systems, users of workstation-class applications, and avid 3-D gamers. Next month, I'll go into more depth about the kinds of applications for which Pentium III will provide a big boost -- and the prospects for breakthrough applications that could make everyone want this power. Regardless, Pentium III's long-term success is assured, even if many users find little value in its higher performance. In 2000, Intel is likely to replace the current Celeron processor with a Pentium III derivative, at which point Pentium II and its derivatives will quickly disappear. People will gladly accept whatever performance level PC makers can deliver at mainstream prices, and in time, the software will emerge that will make that performance valuable to more people. Take me to the Pentium III Special.