Khronos Vulkan API 1.0 release iminant-OSS hetero computing and graphics API
wccftech.com
wccftech.com
I wish I shared your optimism, but without any support on Mac/iOS and Windows support only for desktop apps (I don't think Vulkan will be available for Windows Store apps, though I'd love to be wrong) I don't know how portable Vulkan is going to be in practice.
Once the middleware matures I'm hoping it'll be a good thing for predictability, though. And I'll be surprised and disappointed if tooling to compile DX12 and Metal shaders to SPIR-V doesn't appear in fairly short order, so if you're using an API abstraction layer then apps should be getting more portable.
when I say "much more flexible and portable implementations" I'm mostly talking about GPGPU code.
I hear what you're saying, here's to hoping that SPIR-V will be able to fill in the gaps.
The current article falls in the "announcement of an announcement" category, which is basically always fluff. If the release is imminent, we can have a discussion when it exists.
Actual article title is: "Release of Next Generation, Vulkan API is Imminent – Khronos Group Confirms, API Getting Final Polishes"
These guys can explain it much better than me:
https://www.khronos.org/vulkan
my excited fanboyish rant:
It's an open source cross-platform performance oriented heterogeneous computing API that enables you to build fast low level code that does a good job of supporting massive GPU style concurrency as well as CPU style threads and switch between them at runtime.
With this API it will be easier to write fast parallel code that will make the most of whatever hardware is available while breaking down the barriers that arise when managing threads on both system and GPU memory.
In my opinion it is the foundation for the next generation of HPC. I am interested in this from a performance and from a language perspective. I'd like to see things like PyCuda and PyOpenCL for vulkan and hooks into other higher level languages like Go.
Maybe even whole new languages/runtimes built using the vulkan API as the foundation.
I know I sound a little crazy here but this is something I've been waiting for for many years. Ever since I ran into issues with cuda and opencl.
I would love to see Vulkan take off and get good adoption. Ultimately, though, it's going to come down to how good the driver support is, I think. If there is a serious effort and buy-in from the major vendors and OS providers, then it might take off. Otherwise, it might just be another thing, and contribute to the fragmentation in the GPU space.
Will Apple continue with their Metal platform, instead of, or in addition to Vulkan? Microsoft will back DirectX 12 by default, and if it is a situation where NVidia or AMD or Intel have to provide decent Windows Vulkan drivers, we're back to the same situation OpenGL is in. I don't know enough about the situation in Android-land. General-purpose GPU usage is interesting, but if Vulkan is going to gain mindshare, game developers are the ones that are going to have to pick it over any of the alternatives.
Valve is on board:
http://wccftech.com/vulkan-reason-create-dx12-backend-valves...
Nvidia is on board:
https://www.phoronix.com/scan.php?page=news_item&px=NVIDIA-V...
Google is onboard for Android:
http://arstechnica.com/gadgets/2015/08/android-to-support-vu...
Of course it's too early to know if it will get traction but I am stoked to get a chance to play around with the API. I have a few simulations that no longer work in PyOpenCL and I'm dying to try porting them to something like vulkan.
https://www.youtube.com/watch?v=lnOmy1ly6M0&list=PLCN-Ml6vUJ...
https://code.google.com/p/stableorbit/
But I suspect you'd be more interested in the Github branch if you felt like taking a look at how I and a friend butchered some pyopencl example code and got it working: