I may have missed the beginning of Russian's thing, or I may be reading it wrong, but AMD stated their module design scaled 2 cores to 180% of one core, so shouldn't Bullldozer only have to be ~11% faster to reach an equal Deneb (1 module vs. 2 cores). It's not two 80% cores, it's two 90% cores.
In synthetic benchmarks designed only to test on-chip IO and/or to maximize the use of functional units
(which no useful program will ever do), maybe
(OTOH, such benchmarks, based on assumptions which are obsolete, can show a new CPU in worse light than actual programs).
However, let us look back at Netburst. Throughput, which is what AMD claims to use as a baseline, was improved by leaps and bounds over P6. Any measure of it you wanted to name, being a vague term, was up by amounts that, on paper, looked staggering.
Meanwhile, 1.2GHz P3s and Athlons had no problems competing with 1.6-1.8GHz P4 CPUs, the RAM was cheaper for the other two, and cooling was easier for the P3.
180% throughput can be true, by many measures, and at the same time, completely useless. The more useful one that AMD has bandied about was +80% performance for +50% die area, and even that can't be used to compare to PhII, only to a phantom no-SMT CMP BD. If BD is a Barcelona repeat, all we could end up confident in would be that a pure CMP version would consume more power for the same performance. Quite the consolation, should the performance be terrible, don't you think?
Right, but keep in mind SB is still about 50% faster in IPC than Phenom II is. So even comparing 1 core to 1 core, even if we discount the 80% "penalty" for 1 core, BD still needs to come up with 40-50% more performance PER core than Phenom II does to match SB at the same clock speed.
Turbo boost increases clock speeds, not IPC, and that is why programs with only a few threads are
so much faster on SB, and programs with more thread are
moderately faster on SB. SB only very rarely shows IPC of anywhere near +50% over PhII. AMD has a mountain to climb, but not
that high of a mountain. Ironically, when the extra cores can't help AMD, it is actually TLP of Intel's CPUs giving them such an edge, despite AMD's consistent stance against HT
(FYI: I don't hate HT, I just hate that it cannot be toggled during runtime, so that one could easily test with it, and disable it if it does not suit their needs, without the risk of rebooting from a remote location, or having to wait until after hours).