That’s fantastic news at least in regard to the community’s ability to fix the problem. That is a component we have access to and can improve. Khronos put out a reference compiler, but I’m sure it can be improved upon. It’s not necessarily your job to fix it, but it’d be cool if you did. 
To throw in my 2 pennies, my desktop uses win 7 only because I game. Even then, I’m trying to find the time to switch to Linux so I can get away from Windows. I’m not a fan at all and Ive switched to using Linux completely on my laptop rather than upgrade to 10. I don’t plan on upgrading to 10 and I would be pretty sad to not be capable of playing BS because it’s 10 only.
We’d be delighted to fix it if somebody wants to give us the money to hire another couple of programmers
. Unfortunately due to a lack of resources I’m reduced to just lamenting the current situation.
If you only use your Win7 partition for games what’s the difference between only using Win7 for games and only using Win10 for games?
Just out of curiosity, can any of the issues be worked around by writing cryptic/optimized GLSL that gets the performance you want? Or is GLSL itself fundamentally incapable of expressing the efficient constructs you need?
GLSL is fine and we could achieve comparable performance if we hand tuned all of our shaders to the point where the generated assembly was 1:1. An intermediate idea that we’ve discussed is simply writing an HLSL DXBC to SPIR-V translator which would allow us to use the HLSL compiler for its optimizations.
I’ll admit that’s a fair point and the best answer I can give is an emotional kneejerk reaction away from upgrading.
As is, I’m 100% on Windows for my desktop so it feels like setting up a new OS which I don’t want to/have the time to do right now. In the future, it will probably be a partition only and I may be more inclined to consider using 10 on it.
This idea has been tossed around a lot by other developers. (Well, it’s been tossed around outwardly by people in the Vulkan subreddit. I don’t know if any actual progress is being made.) I’ve never worked in DXBC, so I don’t know how well that would map to SPIR-V.
The decision to leave the shader code optimization in the hands of the community is probably the best and worst idea for Vulkan at the same time. Best, because it means optimizations are now independent from the drivers, so there’s a tremendous potential to see better optimizations ( maybe even faster then DX12 ) in the long term. Worst, because, seriously, until we get to DX12-level optimization, what company really wants to ship their game with a 30% framerate hit off the go ? Especially since the game becomes more likely to be gpu/fill-rate bound ( DX12 and Vulkan reduce cpu overhead, so it’s more likely to get GPU-bound ) and resolution requirements keep increasing ( welcome to UHD and 2-eyes rendering for VR )…
Because you assume everyone else is running on windows 10? Or even be running windows 10 in the next year or two?
I’m honestly a bit perplexed. Complaining about a graphics API that was released one month ago about lack of optimized support. There are a lot of people that are working on maturing the infrastructure, and a lot of companies that are fully committed to it. The only company I don’t see supporting it is Apple
Most likely what will happen is either Apple decides to stop being stupid (not very likely); or some third party stuff will pop up to make metal bindings to vulkan.
Umm, you mean that’s the highest praise you’re willing to acknowledge, right? Because Windows 10 has a multitude of advantages over Windows 7. Major improvements were made to Explorer, proper multi-monitor support was added, first-party virtual desktops were included, and the notification centre is a much welcomed addition. And those are just the simple, off-the-top-of-my-head things that have really helped my work flow, and which I deeply miss while at the office and forced to use Windows 7.
And that’s without looking under the hood.
Maybe those are features that you don’t care about, but going around pretending there are no legitimate reasons for anyone to prefer W10 to anything else is disingenuous at best.
Face it: From a gaming perspective, D3D12 vs Vulkan is going to end up being a significant debate. It’s not a clear cut issue, and there’s no obvious winner at this point. If D3D12 is easier to use, or if it wins out on performance in the long run, developers will chose it. Especially if the linux market continues to be fringe, at best.
And the linux market does continue to be fringe, at best.
Most people haven’t rejected Windows 10 for Windows 7, or linux. They’ve rejected Windows 10 for their iPhones and their PS4s. So if there’s a serious business discussion to be had about which platforms to support, it’s not going to be over focused on the desktop.
Is this tidbit coming to us via the same people that suspected it will use Android?
I believe I also read at some point that the rumor is it will use Vulkan. I haven’t heard anything about it using Android though honestly I’m very worried about the next Nintendo. After some initial excitement everything I’ve read since makes it sound like there’s a good chance it could be another Wii U dud. We’ll find out at E3 in a few months.
HA! I’ve never heard that theory. People think it will actually use Android? XD
No, this comes from the fact that Nintendo quietly joined Khronos midway through Vulkan development. We know Nintendo is not a fan of OpenGL, so we’re all left wondering why they joined.
By the way MoltenVK is likely going to cost a licensing fee, they already charge for MoltenGL, so I don’t really consider that a viable alternative for OSX and iOS.
I know. I’m more curious as to what kind of performance it achieves. We can all sort of sit back and watch them attempt it before investing in some kind of libre solution. Does anyone know how well MoltenGL performs?
Yup! It’s not… completely… bonker… OUYA!
Oh. Right.
It’s just that Vulkan makes a ton of sense on the NX if one believes Nintendo will use Linux or Android as the basis for their OS, and sounds exceedingly un-Nintendo like (standards? What are those?) otherwise.
We can always hope for change, though. Just like we can hope for Wii U sales to pick up! Right?..
I’m sure you’ve seen this:
or
Unreal Engine guys seem optimistic about Vulcan.
That’s because they have the resources to wait until the shader compiler situation gets better or even fix it themselves if they so desire
. I would feel the same way if we were Epic.
My understanding isn’t that Nintendo (or Sony or anyone else) hates standards; it’s that the standards never solved the problems they had. Ever since Vulkan came out, I’ve learned a lot about how graphics hardware actually works, and I’m left wondering how OpenGL ever worked at all. The poor driver authors have to write inordinate amounts of code that guess what needs to happen to run optimally (in addition to all the resource tracking, validation, etc.).
Vulkan changes the status quo by being an API the describes the render process ahead of time in its entirety. So, any given hardware/driver is able to look at the pipeline, make decisions about how to run it optimally, and prepare a corresponding native pipeline with (hopefully) zero guesses/heuristics. That’s what gives me hope about Nintendo and other hardware makers adopting standardized APIs for once in their lives.