Well, what AMD _really_ needs is another A64, that would turn their CPU biz around - it's just not going to happen on a shoestring budget at this point in x86 development.
Which was my point, A64 was a one off because of Intel's desire of chasing clockspeed numbers and a particularly bad interaction of "10GHz-Netburst" with their 90nm process that resulted in Prescott, that's not going to happen again and definitely not on their limited budget this time. Let's be thankful they pulled the plug on Tejas and brought out Conroe. Athlon was great, Athlon 64 was even better, then we got downhill with Phenom. I'm hoping for a competitive product like Phenom II, that's all.
Still, Haswell level performance out of that budget, at a reasonable number of cores (8 probably, especially at 14/16nm) is a force to be reckoned with (5960x for example). That's just a part of the picture though, you've got to consider platform, uncore (cache performance in particular has been a low point for a lot of years for AMD), power consumption, execution as Arachnotronic said... it's a lot of things to cover.
It's not going to be easy for them but if we can even begin to believe they're at such performance levels, it's off on the right foot.
Maybe.
I'm on the same line of thought as Shintai and others on this, I won't believe any of it until the benchmarks prove it. I can believe ATI's side of the hype as that half of the company more or less delivers (more than less I'd say), but not AMD's. Not after Bulldozer. Excavator is actually all in all impressive considering where the architecture started from, but it's the same as Netburst's 65nm shrink, irrelevant.
Zen is supposedly a clean sheet design, so if competent people designed it (Keller and his team), there should be competent results at the end... but a lot can go wrong in the path to that end.
I do want to be able to consider AMD again when my 2500k rig either breaks down or doesn't do what I need it to do anymore in the future.
Time will tell.