Dan, Re: The new P4's boot fine, which leaves perfect or corrupts data. The new P4's were pulled, so they aren't perfect - that leaves corrupts data.
The chips were delayed due to a testing anomaly. This does not mean that there was a problem with the CPU. There are multiple different tiers of validation, and testing applications under Windows or Linux is just one tier. Many levels of testing include custom shells, or even no shell at all - just random seed generation. The validation environment is huge for modern CPUs, and just because there is a failure, it doesn't mean the part is at fault. It could be a test bench issue, a problem with the test equipment, or another environmental issue. Or, it could simply be a transient problem with the board that only surfaced with the newest CPU stepping. It could be anything, Dan, so you are not correct in saying that the only answer is data corruption.