No, we're not AMD fanboys, but NVIDIA needs to get with the times as their Vulkan and DX12 performance shows. To put it simply, Vulkan and DX12 are designed around parallel processing, which is far more efficient. AMD's GCN architecture is designed this way. NVIDIA's GeForce architecture is not, it is sequential and linear thus limiting for games. Vulkan and DX12 can have some stupidly good performance and/or gameplay improvements, but for that we need parallel GPU architecture to be the standard. Someone wake NVIDIA up.
NVIDIA is no stranger to parallel computing and compute performance. See their Quadro cards or even past GeForce like Fermi. Volta should be what we need, but NVIDIA is clearly waiting until the last minute, encouraging the use of primitive and very limiting graphics APIs until then, and slowing down engine development (e.g., Epic Games is too scared to take the next step in Vulkan and DX12 implementation with Unreal Engine 4 since they don't want to leave NVIDIA behind).
Vulkan/DX12 can lower CPU bottlenecks, driver overhead, have epic multi-GPU scaling (including using both NVIDIA and AMD GPUs together), and of course the draw call potential can lead to epic gameplay improvements. In a few years, when Vulkan and DX12 take over as primary graphics API, we will see some very nice improvements in games. It's a shame it can't happen now on a large scale.
This brings up another aspect, and something not as specific to NVIDIA: Multi-GPU technologies, like SLI and CrossFire. These technologies have advanced poorly and NVIDIA is actually regressing and limiting their cards to 2-way SLI only. With the potential Vulkan and DX12 have for multi-GPU improvements, everyone including both NVIDIA and AMD and also game studios should be supporting it.
On a side note, NVIDIA's continued push for exclusivity and proprietary technologies is only limiting themselves and damaging the gaming industry. Volta and future NVIDIA GPUs will be more capable in OpenCL. By that time, NVIDIA PhysX should be ported to OpenCL and not be limited to NVIDIA GPUs. Physics are still in the stone age, with 2004's Half-Life 2 having more realistic and simply more physics than most modern games. PhysX is incredibly realistic in all areas, and can bring massive performance and gameplay improvements if it wasn't limited to CPUs and NVIDIA GPUs.
Thankfully, PhysX has since been ported to DirectCompute, showing performance improvements over CUDA even (which is bizarre). Hopefully OpenCL is next, and hopefully we see the growth in game physics that the industry so desperately needs.
As a side note, it would be nice if high quality shrouds using neutral colors and RGB LEDs, such as this EVGA GTX 1080 FTW, would become a standard design for high end GPUs so that they can match with any PC.
But NVIDIA's exclusivity won't change any time soon. At least their GPU architecture will presumably cease being so terribly outdated and limiting in 2018 with the release of Volta. Of all the hardware industries we have talked about in this article, this one has the most promise of catching up.