Re: Lead time on new CPUs?
I think based on information to date that people *might* want to be asking about "lead time on OLD CPUs" (ie ones that don't have speculative/out of order execution).
If people's systems can't cope without speculative/OoO execution, perhaps people might want to look into the benefits of older simpler software, that wasn't so bloated with shininess as to need the joys of inadequately tested processors underneath.
Any experts on "product liability" laws care to comment here? It would seem unfortunate if the manufacturer of the defective products is the one who ultimately benefits because their customers have to pay again for hardware that wasn't fit for purpose when sold, because it wasn't properly architected, wasn't properly implemented, wasn't properly tested.
And yes I have worked with a team validating an out-of-order microprocessor implemetation. It didn't need to do OoO but having it OoO made it shinier. It also made it less likely to be implemented correctly and less likely to be properly tested.
And look where that leads.