Hope things are well there. I just finished up putting the benchmark through it's paces.
I'll give you the link to the spreadsheet first, then I'll give my thoughts. I did include a couple of runs with two cards OC'd. I also included one run with two cards with motion blur disabled.
I basically used stock configurations for all four cards(GTX970/GTX1060/RX470/RX480). The only adjustments to the cards were the fan profiles and the power limit. I set the fans static to 65%. I did add enough power to the AMD cards to make sure they maintained their stock settings. The GTX1060 was set to 116%. The 1060's are a bit under-powered in my opinion. The GTX970 is on my own custom bios, so the only adjustment to that card was the fan profile(65%). My CPU was also set to my everyday OC of 4.0ghz.
I'm not sure where to start in discussing the actual benchmark. I guess I will start with the positives and the technical.
I only had the benchmark crash once. I'm guessing I did probably about 40+ runs total. The crash wasn't due to the benchmark, it was me setting the OC too high messing around. lol
The benchmark does such a better job at keeping a constant core load than Valley/Heaven.
There seems to be a bit of difference in the amount of Vram used in-between AMD/nVidia. It seemed at the same settings the AMD cards were always using a bit more Vram. This could be because of their uArch or the fact the A-Sync is baked into the hardware. I'm not sure if this uses A-Sync or not, just a wild guess.(I think that is a feature of DX12 now that I think about it)
The benchmark picked-up the core and memory speeds perfectly, even on the Pascal card. Pascals chips do have a non-static clock. This is due to nVidia's Boost 3.0(FYI, Hate it!)
Vram usage that I touched on above. Again, that issue may be hardware bound, so nothing Unigine can really do.
The disparity between the two mfgs. Like Valley/Heaven the benchmark "appears" to favor nVidia. This has been discussed in-depth across the web. My personal opinion is, "It's just the engine." I don't think it is done on purpose. It's just like some games run better on one card than they do the other.
The benchmark is single-thread bound.
I am a bit disappointed in this really. On the other hand, from what I have read, this benchmark will support DX12. Correct? If so, then we may see a world of difference in the results. I think it is widely known the AMD holds a bit of a edge in DX12. I can understand going single thread, it isolates the GPU. It is a fine line that you(developers) have to walk.
I like it! It's not a walk in the park like Valley/Heaven is for today's GPU's. It really makes the GPU work, and that's a good thing. It looks amazing! It has sort of a old fashion feel to it....like the 40's(WWII) to me. I love all the information that it is giving the user too. I'm a bit of a numbers junkie. I think that comes from my love of baseball, and all its stats.