Popular wisdom has P5+ pegged at 2.4GHz, a number that would put the final nail in the IPF coffin while filling up the IBM coffers. The problem is IBM won't get there at first, not even close. In initial contract bidding, the firm won't put forth anything more than a 2.1GHz model no matter how nicely you ask. When asked, sources said heat was the culprit here, which is odd considering that IBM stated that power management was a key goal of P5+.
Either way, it looks like the 10% or so speed bump will be more than enough to keep them ahead of IPF for the next year or so. If the problems are real, that should be enough time to make the necessary tweaks. If they are sandbagging, well, IPF can't be any deader.
Now, this whole situation is bad for the business side of the IPF market - they are far less bandwidth bound than HPC, and way behind on TPC-C already. In the HPC markets, the fight is closer, so it will be less of a problem, but still a headache.
To make matters more confusing, we hear that the silicon for P5+ is in amazingly good shape, while others say that it's screwed up in a serious way. Nothing definite other than there could be just about any reason for the 2.1GHz launch. ยต