There are certain physical laws that sometimes limit what one can do. At least, that's true until technology finds a solution to what had been a conundrum. A good example is the so-called "sound barrier." The "sound barrier" turned out to be no barrier at all once there was sufficient power available in aircraft engines, and airframe designs had been modified to handle the buffeting at trans-sonic speeds.
But in processor performance, we're coming up to a decision point. As processors get faster and faster, the heat they generate is making other alternatives seem more attractive.
Power dissipation in processors varies directly with frequency: As frequency goes up, so does the power the chip uses, and that means it gets hotter. So if you want to increase performance, heat becomes more of a problem. "We can get some performance gains with increasing clock speeds," says Intel spokesman Howard High, "but there are some other engineering challenges. As frequency goes up, so does heat. You have to manage it."
Well, users, and that means IT folks who want their servers to go faster, demand more performance. So both Intel and AMD are now talking about making dual-core X86-code-based processors that will boost performance at nominal clock-speed increases, rather than relying only on higher clock speeds.
"If you can end up with two processors instead of one in each CPU socket, then you can get the ability to perform tasks in less time," says Barry Crume, a director in the server and workstation segment for AMD.