2008 8 core Mac Pro or 2009 Quad

[Transporteur]

Doesn't this prove the exact opposite? The 2.26 has 16 effective cores and the 2.8 has 8 effective. So if Hyperthreading is working as expected and scales correctly the 2.26 would be much faster even with the 600MHz difference (In geekbench land at least) It has twice the cores and geekbench is using them. This in no way reflects real world of course. For reference my 3.33GHz 6-core benches 14334 on 32-bit and it would destroy the 2.26GHz on pretty much everything. I see what you are saying but HT does work with geekbench and scales better than most other complete core saturation apps aside from Cinebench or something.

No, the machine does not have 16 cores, it still has 8 physical ones. Hyperthreading just emulates cores which gives you another 10 to 20%, but not a 100%, which would mean that you'd have 8 physical cores + 8 logical that are exactly as fast as the physical ones. No es possible.

 

[derbothaus]

That's why I used the word "effective". I did not specify "physical". On highly multi-threaded apps you get way more than 10-20%, my friend. Closer to 60-80%. Just take a look at Cinebench i7 numbers of any flavor.
For a test I ran the Logic Benchmark from macperformanceguide,
65 Tracks with 6-cores. HT disabled.
119 Tracks with HT enabled. I call that way more real world and way more than 20%.