Have you ever played CIV? It isn't a graphics intesive game, it's a CPU one. Nvidia will be fine. Especially since looking at the graphics for civ 6, they are cartoony and look like they won't exactly take a lot to run.

The graphics have never really been the issue with Civ games. The amount of draw calls has been. You have a lot of CPU work for AI and the rest, and a TON of draw calls for later parts of the game. It's an ideal candidate for DX12/Vulkan. Beyond Earth was quite amazing with Mantle.

Why all the talk of drivers like this is like older api's? They are not totally irrelevant, but no one is likely to release a game changing driver that will fix a whole api. If a game is not using Nvidia GPU's as well as it should, then the devs will be the ones to fix.

NVIDIA's architecture is way to graphics workload-focused, and much less parallel than AMD's, with much worse hardware scheduling. In a sense, it's still stuck in the older paradigm of making games where a single core was submitting work to the GPU. Admittedly, their cards work wonders with it (100% utilization), but don't expect that kind of design to benefit from newer APIs, and don't blame the developers for it.

Civilization V, and I guess the next VI, is a lot about memory management. Graphics API are not only about graphics effects and draw calls.. Moreover Civilization games are meant to be player even with low-level GPUs (CIV V had a patch to add touch-screen support too), so everything that will improve graphics performance will be appreciated by a lot of people. Also on mobile the reduce of power-requirement due lower CPU overhead will be a lot appreciated.

Civilization V, and I guess the next VI, is a lot about memory management. Graphics API are not only about graphics effects and draw calls.. Moreover Civilization games are meant to be player even with low-level GPUs (CIV V had a patch to add touch-screen support too), so everything that will improve graphics performance will be appreciated by a lot of people. Also on mobile the reduce of power-requirement due lower CPU overhead will be a lot appreciated.

Wouldn't a Vulkan (instead of DX12) port make more sense for Civ? I mean, Vulkan seems to be the de facto API for Android. And it runs everywhere else.

It probably comes down to what the devs are most comfortable using, they/AMD may have also prioritised explicit multi adapter and think Vulkan is a bit behind on that front.

i'm all about dx12 and Vulkan replace the aging DX11, it just seems nvidia arch can't adapt to dx12 as well as AMD, which means AMD gonna have a good gains while nvidia won't have any, like all DX12 games now.

nVidia's architecture is fast and simple. AMD's is more robust and complex. Fast and simple was good in past as workloads were simple. For future (I guess everyone wants) projects will be very complex with more stuff on screen, more stuff in background. And as those complex workloads come through proper API (DX12/Vulkan) AMD's complex HW becomes suddenly better utilized than before and able to deliver all it could not do for those older simple workloads. But take nVidia's fast and simple GPU and push all that chaos on it and suddenly it may get overwhelmed. - - - - Now, I do not think that throwing a lot of stuff on any GPU is good idea, especially if some kind of object culling can be implemented on engine side. But in future we will see many more games with high amount of small objects/local effects instead of wasting processing power on few very detailed ones. My guess is that in year's time we will see really nice clutter on ground in adventure games. When I look at most current games, you can go through garage, and it looks quite tidy, nearly unused. But that's because no developer wants to spend time on creating details he knows will negatively affect performance. - - - - DX12/Vulkan simply increases spending budget for many things and therefore it is quite logical to expect that some developers will do more in their projects than before.

My guess is that in year's time we will see really nice clutter on ground in adventure games. When I look at most current games, you can go through garage, and it looks quite tidy, nearly unused. But that's because no developer wants to spend time on creating details he knows will negatively affect performance.

No developer management wants its developers to spend time. Period. Because time is money. Even with yesterday's GPUs games could have had a helluva lot more clutter around, but it takes time to model and texture it and finally put it in the maps. Especially if you want to avoid the feeling the game character owns 100 identical wrenches, placed all over his garage. I'm no game developer but in the past I spent countless hours modding games, making maps, so I know it from experience.

nVidia's architecture is fast and simple. AMD's is more robust and complex. Fast and simple was good in past as workloads were simple. For future (I guess everyone wants) projects will be very complex with more stuff on screen, more stuff in background. And as those complex workloads come through proper API (DX12/Vulkan) AMD's complex HW becomes suddenly better utilized than before and able to deliver all it could not do for those older simple workloads.

Slide 39 http://gpuopen.com/wp-content/uploads/2016/03/Practical_DX12_Programming_Model_and_Hardware_Capabilities.pdf Slide 41 (GCN 1.0 is like GCN 1.1 except for for tiled resources tier1 wich mades them FL 11_1) http://gpuopen.com/wp-content/uploads/2016/03/Practical_DX12_Programming_Model_and_Hardware_Capabilities.pdf GCN is soooo complex that is has less constraints about memory management and resource binding (mostly: constant buffers) and compute shader usage. What GCN really lacks is a modern and advanced rasterizer.

But take nVidia's fast and simple GPU and push all that chaos on it and suddenly it may get overwhelmed.

Huh! Do you see that happening with the current generation GPU's (Pascal)? As I recall AMD's newest generation GPU still doesn't support feature level 12_1 ... where is the future proof in that?