Intel talked about their future plans the other day, and announced that between now and 2010, they will introduce two new microarchitectures, Nehalem in 2008 and Gesher in 2010.
I can hear the yawns already, but really, this is a good deal more important than you might think. This fixes Intel’s biggest problem.
What has been Intel’s greatest failure the last half-dozen years? They’ve run CPU designs into the ground. They pushed them until they broke.
They pushed the PIII design until it broke. They took years to get out a PIV design which gave almost immediate signs of terminal current leakage. By the time the PIV hit 2GHz, 40% of CPU power was leaking away. They kept on, hoping to find enough fixes to keep things going, and they ran out with Prescott.
And why did they do that? They had one core design meant to be stretched for generations of processors. Well, it didn’t. It didn’t twice.
Coming up with a new design every two years is no guarantee that they’ll be good designs, but at least if Intel rolls craps again, it’s not like they have to wait three years to really fix it.