So does Vulkan lock people out or something? From what I'm reading of what you said, it sounds like there will just have to come a point where some expansion will create a fork, if the demographics show it's reasonable.
It reminds me of that tumultuous time between 32 bit and 64-bit gaming, before everything really switched over. At some point, you just have to make the move.
The business question is, will you attract more players with a better quality product than having technology friendly to older machines?
But that is far down the road, I suppose. Who knows, maybe your modding community will help push that issue for you with what comes out.
Vulkan doesn't block anyone per-se, but older hardware doesn't support the basics of it. Below OpenGL or DirectX, there's a thing called the "shader model" that graphics cards support. When graphics hardware says it supports "directx 11," it really means that it supports a certain underlying shader model (well, and presumably that it has directx 11 drivers, but that's software).
OpenGL and DirectX are both a layer up from the drivers, and the drivers are simply a method for interacting with the underlying shader pipelines in specific ways. Vulkan is basically "new OpenGL" (which they thought about calling something along those lines, but didn't because reasons), and it's a lot like DirectX11 in terms of the raw feature support (tessellation, etc)... except it's unlike
anything else on the scene right now. I can't really say "the market," because it's not something that's sold. It's like HTML5, in that it is kind of a standard that a lot of industry titans got together and designed to solve their problems.
One of the problems that is largest has typically been that there's a lot of load on the CPU as it spends time preparing to push stuff into the GPU pipeline, and there's also a lot of missed chances where the GPU is actually doing very little because it's waiting for the next instructions that have not come yet because the CPU was waiting to hear back the GPU was done before it puts the next thing on the bus... etc etc etc.
So Vulkan comes in and basically says "we're going to multithread the heck out of this," and also "we're going to keep that pipeline full, so your GPU isn't ever just sitting around waiting for instructions." Those are huge improvements, because not only does it free up the CPU some, it's actually a haha-bad-science-you-only-use-10%-of-your-brain type of unlocking of the GPU's potential. Depending on the moment, we only are using 40% of our GPUs (I guess?) with DX and OpenGL at the moment. It sounds ludicrous, but that's the nature of a bottleneck or a race condition; if nobody ever tapped their brakes and drove with machine-like efficiency, we could get twice as many cars through every city with no traffic jams. Or something like that. That's probably an exaggeration.
But anyway, Vulkan is aimed at breaking down the traffic jams in much the same way. So it makes older hardware "seem better," and in general able to do more when it comes to a variety of things. If you were already struggling with something like pixel fill rate, then you weren't bottlenecked in the first place: that's an example of the highway simply being too small for the number of cars, and no matter how efficiently the cars drive you still can't get any more in there.
So it's not some sort of magic bullet that solves every problem. But it does solve a lot of the problems that are most problem, and overall your GPUs will run a lot more thoroughly (and possibly hotter? hmm) doing whatever it is they can do.
THAT said, it's also a cross-platform way of handling a lot of the DX11 style features, some of which were spotty in OpenGL before. Things like tessellation were hard to do consistently between the two platforms, and unity doesn't even support tessellation in openGL.
More relevantly for Arcen, it also makes available GPU instancing, which is the ability to send one mesh+material and say "draw this n times at this list of n coordinates." As you can imagine, this is a lot more efficient on the bus between the CPU and the GPU, and it also allows the internal GPU drivers to reuse some work to my understanding. That last bit is more of a black box, but anyhow. It basically lets you do something like draw 40k asteroids in the time it might have taken to draw... I dunno, maybe 100 before. It really depends on your hardware, again, and what the limiting factor is. There's no one way to quantify it, which is why you don't see generalized quantifications out there on the internet. If fill rate is your issue and your GPU only has so many cores and so on, then maybe those 40K asteroids don't actually render so insanely much faster. Faster, sure, but not as much more as it could have been. If your card is more modern, though, then the pipeline was almost certainly your limiting factor, and as time goes on that was only going to become more and more of an issue.
SO.
Does Vulkan lock people out? Not really. But what do we DO with all that power?
1. We use it, and rely on its performance gains in order to make the game acceptably playable framerate-wise as well as looking great.
-- Result: people not using Vulkan get terrible performance by comparison, possibly making it feel unplayable without a huge amount of stuff turned off. Also it's more work for us to figure out how much stuff to turn off to let them play in this stripped-down fashion.
2. We use it, but just to give people insane bonuses to speed over the "baseline" speed that is more rooted in DX9. So anyone who can use it has performance that just fliiiiies, but people who can't use it still get an acceptably platable framerate.
-- Overall this is my intended approach with this particular game, at this time.
However, either an upgrade in a future version of unity, or some change that we decide upon ourselves, might suddenly make #1 a lot more attractive or even "required in order to move forward in a modern way." We're talking years from now, most likely. So the question then becomes what happens to people with 2012 or 2013 cards in 2019.
