CPU designers beg, borrow, and steal for every ounce of performance they can get
Alan Joch, Senior Editor
So many factors go into making a great computer. You need to consider I/O, storage, memory, and the display. Oh, and don't forget ergonomics; human factors are very important. But what
really
gets your heart pumping is what lies at the heart of today's best systems: the newest, most screaming, most radically designed CPUs.
We have become addicted to speed. Gordon Moore is our pusher. Moore's law, which states that processing power will double every year and a half, has thus far held true. CPU designers, always in search of a better fix, drain every possible ounce of fat from processor cores, squeeze clock cycles, and cram components into smaller and smaller dies.
They also steal ideas from each other. As "
CPU Scorecards" shows, the line between CISC and RISC is getting blurrier every day, and the next-generation offerings from the major chip families continue this trend. The upside for us is that a SPECint95 score of 300 will soon be the high-performance midrange, and we'll be eyeing scores of over 500 by this time next year.
However, processors aren't just getting faster; they are becoming more specialized as well. Consider the handful of new multimedia chips that we describe in detail in "Chip Fashion."
Perhaps nothing epitomizes the merging of CISC and RISC and the rise of specialization more than IBM's stealth processor, the PowerPC 615. This single-chip marriage of PowerPC and x86 promises to run native applications from both platforms with little performance hit. Insiders say that IBM has been working on it for years and has found the design mind-numbingly difficult to build.
The latest round of rumors puts the chip's introduction sometime in 1996. IBM, which barely admits that the proj
ect even exists, won't talk technical details. So, we asked a leading expert on chip design, the
Microprocessor Report
's Linley Gwennap, to design the 615 for us. As far as we know, this design, outlined in "Why the 615 Matters," isn't necessarily what IBM is working on, but the efficiencies of Gwennap's shared-memory model may be fodder for Big Blue's engineers.
Whether you admire CPUs inherently or just appreciate them as one of the many important system components, the technologies described in the following pages should raise your pulse.
Heart Throbs
