Heh... so in an effort to prove myself correct, I've spent the last 90 minutes benchmarking my i7-2600K @ 4.3GHz against itself with hyperthreading turned on and off.
For the sorts of stuff I do (photoshop and x264 encoding) I'm seeing a gain of about 20% when using the CPU in hyperthreaded mode. On something like first pass encode, I'm seeing hyperthreading around 10% slower than when it is switched off. The code is obviously spending more time managing the switching than doing the work.
That's a lot less than I was expecting. Just as well I am hungry - the humble pie beckons....
