Thanks, that was very interesting for me.
Compared to the benchmarks I have been getting from my early 2009 2.66 Quad I am seeing an 11% gain for single CPU and 45% gain for Multi CPU.
For all the critics of the new Mac Pro range, I think that is a pretty good increase at the entry level base model price bracket. Ok the 11% is a little weak, but the clock speed is lower. I think the multi CPU results is a great improvement on an already impressive machine.
Cheers
Cris.
I'm not doubting your scores, but something seems odd.
So 11% faster than the 2.8ghz Quad, I am assuming you got around 3570 for the single CPU and at 45% faster for multiple CPU you got around 14870.
Those scores seem to be about the same as what I've seen floating around.
Now the thing I find odd is how multiple CPU's have actually given you more than 4 times the single CPU score. Even at 100% efficiency with no overhead, you should be getting no more than 14280. You're actually getting a multicore speedup of about 4.16x. That seems... impossible to me.
Actually, maybe not. Yours uses hyperthreading, right? So maybe what we're seeing here is hypterthreading actually being beneficial. But it still doesn't makes sense, because I thought hyperthreading was only useful if the core wasn't 100% busy. If 4 cores are 100% busy, the hyperthreaded cores (not real cores) should also be 100% busy.
What really strikes me as odd is only the 2009 2.66ghz Quad seems to be getting this time of speedup. The 2009 Octo's with Hyperthreading are still only getting about a 6.2-6.3x speedup. The 2009 2.93ghz Quad gets a 3.76x speedup.
So I think some test here is wrong.
Or maybe this has to do with Hyperthreading, or maybe how Cinebench is working and how many cores Cinebench is really using.
We need more samples I think.
