Hmmm...
First, AMD didn't invent Hypertransport, It is Digital, maker of Alpha Chip. Hypertransport is no BS technology like MMX.
Second, Intel didn't invent Hyperthreading, It is Digital again, maker of Alpha Chip. Hyperthreading does not taking any two thread into the CPU like SMP.
Alpha Chip is 10 yr advance in the time when it was introduced. The Digital Engineers are smart, however, the chip was too advance and the industry was not ready for them. Marketting failure (like many great tech companies). Compaq bought Digital, however the smart engineers of Digital went to different companies. Some went to Intel, some went to AMD and some went to IBM and Motorola.
CPU has two calculation units. Integer and Floating point. CPU w/o hyperthreading can only do one type calculation at one time, but CPU with hyperthreading can do 1 integer calculation and 1 floating point at the same time. Hyperthreading would kick off if the application is optimized for hyperthreading. However, the real world's performance of hyperthreading cannot be measure by benchmarks. Hyperthreading is fast. Not faster than SMP, but definately it is fast.
Then, what happened to those engineers who went to IBM and Motorola? their idea is multi-die on a single CPU, making a single CPU act like a quad CPU. (and they have not yet succeeded) They saw the limitation of hyperthreading as it requires programmers to optimize their application to take the advantage of HT. How many programmer will actually busting their *** just to take the full advantage of hyperthreading? Not many, sure they all going to optimize for hyperthreading , but fully optimized for HT? I am very doubtful. So that is why you will not see a boost of 50% performance gain but a moderate performance gain like 10%~15%