I don't think I've ever heard anyone say that it wasn't okay to compare an i3 (for example) to an athlon II x4. . . As far as I'm concerned it's okay to compare any cpu with any other cpu...but for the comparison to actually be valid concerning purchases, we need to compare similar platform costs.... Ie the i5 750 vs x6 1055 is a comparison that makes sense, i3 530 vs athlon II x4 is a comparison that makes sense, i3 530 vs i7 980x...doesn't make any sense...Shrug.
Anyway...so as far as threads go. I wonder if AMD has any plans to introduce hyperthreading type features? I know that bulldozer supposedly has a lot of components shared between the INT cores, but not to the HT level, there are still two execution units. AMD has smart engineers and I'm sure they'd used it if it enabled a huge performance boost...so wonder why they don't?
Also curious about other architectures. What about ARM, for example? Isn't HT a very power efficient way to increase performance (on Intel stuff anyway)? Do you guys think that it will spill over (SMT, not hyperthreading specifically) to ARM and such? Or maybe ARM already has something like it?
Then I kind of wonder about taking it one step further:
1. Some workflows don't benefit from HT, and in fact may be hurt by it...
2. Would it be possible for the processor to look at the thread that is executing and somehow "check" the execution units and enable or disable HT as needed?
3. What about if it was something the processor detected might benefit from more than one more thread? Could we turn on maybe 3 or 4 threads per core if needed, and disable them automatically if not? *shrug*.