Bulldozer is a streamlined design, that is the whole point of the exercise, their dudes came to the conclusions that a lot of the time the current design has resources sitting around idle and doing nothing, therefore wasting space, power and heat. so in Bulldozer everything is leaner, more compact and theoretically more efficient, each core has less resources than the standard cores they use today but run at a substantially higher clock speed with more aggressive pre-fetching and such, so would expect them to be as fast or faster most of the time, in most situations.
would only expect each 'core' to fall behind in situations that use all the resources of conventional designs, the biggest factor with Bulldozer is by redesigning and streamlining they can fit two integer cores and one shared floating point core in each module, which only takes up about 10% more space than a standard design, but has more resources than a standard design and is much more flexible than the previous generations. each core has fewer arithmetic units but since you essentially get two of these 'cores' on in a similar die space to the previous design with a better, more flexible floating point unit your going to see some benefits, especially when in a multi-threaded environment.
so for 10% more die space (for each core) you get 50% more resources (Phenom II has 3 ALU, 3 AGU, Bulldozer has 2 ALU and 2 AGU in each 'core', so at a module level it has 4 ALU and 4 AGU) at a core level, a much higher clock speed whilst still running within the same sort of thermal envelope as the previous cores, not to mention the cores are using more efficient use of these 50% more resources and wasting less time doing nothing, I think Bulldozer is an awesome piece of micro-architecture design and should totally do what they intend it to do. stop comparing it on a core vs. core level, that has never ever been the intention of the exercise, you compare one Bulldozer module to one Intel core/one Phenom II core or else your forgetting it was designed from the ground up to work in that exact sort of competition/environment. so vs. Phenom II is has more resources, can execute double the amount of threads and has a better instruction set running on a smaller process at significantly higher clock speeds whilst using equal/less power, how can that not be considered a success though I am sure the Intel camp will try think of something?
Edit: how can it be disappointing if we get eight higher clocking cores on a marginally bigger piece of silicon than current four core Phenom II, double the cores for like 10% more space clocking higher without using more energy...?
