If you’re looking for AMD to repeat the success of its Athlon64 line had against the Pentium 4’s then you going to be disappointed. AMD’s strategy isn’t to design and manufacture the world’s fastest microprocessors but to sell the most efficient, cost effective and best value for money CPU’s in the world.
Bulldozer has the potential to be a hugely successful CPU, with integrated graphics from ATI it will potentially wipe the floor with Intel’s Core i3’s and dual core i5’s. Architecture improvements and AMD’s keen pricing could also make Intel’s high end enthusiast chips almost redundant for most users.
THere isn't a question Bulldozer will wipe the floor with i3/i5, bulldozer as it is won't have intergrated graphics until probably the second interation of it. Because frankly an 8 core bulldozer will obviously use a lot more transistors and space than a 4core would, so theres space for gpu's to be intergrated alongside a quad core. Theres also little interest, for now, for higher performance graphics in the server market, nor the high end home market. Llano, for now, will be the "intergrated gpu" i3/i5 competitors, and due to the graphics probably being significantly ahead will find its way into many systems for gaming on the cheap(very cheap, lots of people still game(badly) on Intel intergrated though. Llano will be great on price/performance and hugely ahead on graphics performance to go with it.
Bulldozer looks like its going to be VERY competitive with Sandybridge, and I wouldn't be remotely surprised to find it faster in several situations and slower in others. Bulldozer is probably a more long term design than Sandybridge though, AMD have spent money and are bringing in personel to push the intergration of software to use the on die gpu's. Bulldozer is designed with an explosion of FPU power in its second/third iteration as a gpu core is intergrated on die. AMD's designs have generally been FPU heavy and int light in terms of power, while Intel have generally been the reverse of that. Its there supreme interger performance that for a few years now have given that huge performance in lots of things, though a lot of it is simply a new design and not having to deal with older issues.
Bulldozer looks to be the biggest increase in Int performance on an AMD core, really ever, while FPU has taken a less big increase, but is still a very efficient and powerful FPU upgrade in there.
BUlldozer may/may not trump Sandybridge, but it shouldn't need to, Sandybridge is significantly faster than a Nehalem, its NOT a small upgrade at all, its not far off a completely new architecture and is a huge leap forwards, if AMD match that, they can be incredibly happy with it.
THe problem for Intel is when intergrated GPU's become standard, AMD will offer the best Fusion chip, without question and the upside of that for AMD is they will have almost certainly, very good power consumption and performance for the gpu part, meaning more power available for the CPU and a more flexible chip.
IF gpu offloading of data becomes standard, which seems to be something both Intel and AMD are pushing software guys heavily towards, AMD will have a simply huge advantage for a few years at least.
Bulldozer will be great for high end users straight off, I'd guess a year later, when we get intergrated gpu versions, and maybe some quad core versions with intergrated potentially a lot earlier(but will basically make Llano obsolete) will be the really really competitive to Intel chips when AMD can start to make a huge impact. Remember theres also not a huge point in AMD having chips 10 times better than Intel right now, their new production, as in the New York State $8 billion fab, won't be able to pump out huge numbers of chips till late next year/mid the year after.
AMD have every, EVERY intention of trying to make the fastest processors around, suggesting otherwise means you have no idea who owns the company and their goals for it. The reason AMD have been making the chips they have till now, is they've been hampered by many many things, lack of cashflow, lack of places to make chips. IF they'd spent 4 times as much on R&D, and demand went up 4 times, they couldn't produce 4 times more chips, so the chips they made would have to make 4 times more profit to break even. Those problems have quite literally gone, cashflow, not a problem, production, well on route to being a non existant problem to the tune of the biggest and best fab the world has seen being built and that fixes every problem they had.
Phenom 2 is actually a truly fantastic chip considering their budget, design goals, and the limits they had imposed.