Holy crap this is the greatest, and most humorous benchmark presentation in the world!!!1 Go give this a run guys. At worst, go watch a video of it, but really you don't want it to be spoiled beforehand...
Running it a second time while looking at task manager, nah, not really. The CPU Physics Test hit 80% across all eight cores on this FX-8350, but everything else hangs out in the 20-30% range. I'd estimate 40-60% across the four cores of something like the 3570K that everyone around here likes? That leaves it as heavily leaning on the GPU.
Here's me - Nvidia GTX 670 + i5 3570k (no overclocking)
Got killed at one point in the physics stress test (the collapsing building), and the CPU test was hovering around 32fps (the glowing orb or whatever). Otherwise, the rest was well over 60fps for the most part, so I'm pretty happy with that result after not tweaking anything.
"My advice: don't spend money on therapy. Spend it in a record store." – Wim Wenders
Interesting that the CPU Physics Test comes out so much higher on your 4-thread 3570K than my 8-thread FX8350. I'm guessing that it is heavily floating point dependant, or using an Intel-specific instruction. The 3570K does have a general instructions-per-clock advantage, but the extra 600mhz tends to make up for that most of the time. I am missing an FPU on half my cores though. I'm now curious how a 3770k or a six-core Phenom II do on that sub-test.
A GTX670 is roughly twice as fast as a Radeon 6870. The almost direct doubling in score based on very similar CPUs leads me further to believe that this bench is highly GPU dependant.
Digging through the results leaderboard I think I've got the CPU Physics Test figured out. It is still heavily reliant on the GPU and scales almost directly in line with the GPU, though it does like more threads.. A 3570K running onboard video scores in the 185 range, while the same CPU with Daver's GTX670 is in the 700 range. Interestingly, a 3770K with it's extra four threads scales pretty well in to the 1000 range.
We can pretty much count this as solely a GPU bench. It's still be one with the best presentation.
Yes. Nvidia control panel was set to all default settings. So much for it being a heavily GPU based benchmark. The only thing I can think of is, I did not install the other two programs that the installer wanted me to. Oh well, I'll stick to 3DMark and Heaven for my benchmarking needs.