[citation][nom]captaincharisma[/nom]i wonder too. they admitted defeat in the mhz race a few years ago and changed there nameing scheme for there CPUs to show that, the phenom CPU was a big mess out of the gate and now the same thing is happening to the bulldozer. OH! im sorry we will have to wait for a new OS to truly see how BD performs. soory fanboisface it the only success AMD has now is because of ATI. while there dollar store CPUs continue to declineFAIL[/citation]
Well, for what it's worth, AMD should win the MHz race with Bulldozer, based on the design. Of course, they still have to execute on it, but I don't think MHz matters so much anymore. Back in the days when the Coppermine and Athlon ran roughly equally at the same clock speeds, people paid more attention to it. Now, with different IPCs, and weird model numbers, it's not as important a yardstick.
AMD surrendered to Intel in performance when they sold their fabs. Intel can integrate the two in a way AMD now can not. I hate absolutes, so we it's unwise to say never, but the chances of AMD ever outperforming Intel again are very, very slim. People point to the Athlon 64 outperforming Pentium 4, but that's when AMD had fabs, and Intel was selling a processor that was designed without much insight into the power and thermal restrictions that would suffocate it.
All that being what it is, it's not very important. The world doesn't revolve around kiddies shooting space aliens, but does revolve around money. Bobcat is a very good product, and is attractive for a huge market. It's a little heavy on the GPU side, but that's about the only negative you can say about it. It's very small, has decent performance (unlike the Atom), uses low power, and has a high level of integration. They did a great job on the technology, and did a great job in positioning it into a huge market. It's more important in a positive way than the Bulldozer is in a negative way.
Also, there's nothing fundamentally wrong with the Bulldozer's high level design. The implementation obviously leaves a lot to be desired, but I wouldn't be too surprised if the next iteration at least exhibits decent performance. I don't think anyone believes it will ever match Intel in single-threaded performance, but, in certain workloads, assuming a decent improvement in the next iteration, it could be useful. No doubt it's a failure now, but, like the Pentium 4, there are some good ideas there that might show themselves in better implementations.