You may be right about Katmai, especially if Intel markets KNI (a.k.a. MMX2) as well as it marketed MMX.
Should be easy, since unlike MMX, MMX2 is actually useful. :-)
However, despite all the hoopla about IA-64, it's still only going to represent a very small part of Intel's total volume. McKinley, the follow-on to Merced, will probably take off from where Merced started volume-wise, but it's still unclear as to how much revenue the IA-64 line will earn for Intel because of its relatively low volumes.
My thinking on IA64 has evolved over time. The nerd in me is still fascinated by a new and radical architecture, but I wonder how the market will greet it. If it only sells to the high-performance FP crowd, what kind of a success would it be? Not much of one by Intel standards. In fact the only real impetus to switch to something new is the 32-bit addressing limit, which won't hit hard for several years. EEAX, anyone? |
