Yes, they intended for this fail launch

. All that marketing hype, the bringing back the FX name the "AMD unlocked" lark, all leading up to Piledriver, give me a break. I suppose Llano bad yields are also a lead up to Piledriver?
PileDriver isn't going to pull a conroe I don't think, at least given AMD's own performance increase figures which IIRC, are 20% over Zambezi. For all we know, it's all clock speed increase, or all IPC increase, probably a bit of both.
On the flipside, Ivybridge is only going to be a few percent higher clock than SB and a few percent better IPC IIRC, it's not going to be any substantial improvement.
Beating the 8150 depends on the situation.
An E5400 will wipe the floor with the FX 8150 in PS2 emulation for example.. The ever popular Rome Total War will perform better on an E5400 etc.
People have a lot of uses etc, but it's not just the 2500k that beats it, or the Phenom II's at all.
DM cares too much for theory, fair play to him, but knowing the theory is completely different to knowing the practicality. Take BD for example, the modules, I said about two threads being ran off a module would cause a performance hit, DM said I wasn't understanding it right. BD launches? Oh look, the two threads being executed from BD does give a performance loss in scaling between the module. The scheduling "patch" should improve situations in applications that utilise 2-6 threads.
Knowing/Speaking the theory like DM does, gives the advantage of being able to, I suppose, describe what you're saying, which is where I fall down, I can't tell you why something is the way it is, just that it is that way.
Although, the scaling was worse than I'd suggested.