"Vulkan Next" Is In Active Development

[phoronix]

Phoronix: "Vulkan Next" Is In Active Development

The Khronos Group is having their day at SIGGRAPH 2016. No major API announcements with availability today to talk about, at least not yet. But "Vulkan Next" was commented on...

http://www.phoronix.com/scan.php?page=news_item&px=Khronos-Vulkan-Next

[anarki2]

"a rigorous memory model"

Uh, wasn't the whole point of these shiny new APIs the fact that they wanted to get rid of this?

[edoantonioco]

What about better debugging tools, better example codes, mature tools, etc? I have read about people saying that as an argument about why they choose Dx12 instead of Vulkan, and that sucks, I hope than khronos will improve on this point (if they really want than vulkan can compete with dx12 on AAA games). Another thing is than AAA devs with dx12 can have the support from microsoft if they need help, but what on the case of vulkan? posting on a community forum?

Anyway, multiGPU support is something than vulkan is lacking right now compared to dx12, so they have to release it as soon as possible

[Zan Lynx]

"a rigorous memory model"

Uh, wasn't the whole point of these shiny new APIs the fact that they wanted to get rid of this?

I think that they probably mean to make it more explicit what operations come before other operations, so that developers don't fall into the trap of writing something that works on their hardware but not on a phone, for example.

[andre30correia]

What about better debugging tools, better example codes, mature tools, etc? I have read about people saying that as an argument about why they choose Dx12 instead of Vulkan, and that sucks, I hope than khronos will improve on this point (if they really want than vulkan can compete with dx12 on AAA games). Another thing is than AAA devs with dx12 can have the support from microsoft if they need help, but what on the case of vulkan? posting on a community forum?

Anyway, multiGPU support is something than vulkan is lacking right now compared to dx12, so they have to release it as soon as possible

in linux amd crossfire don't work (even in windows it sucks) and sli is become less and less necessary, having a single and powerfull gpu is much better, only 1% of gamers use it

[gufide]

Cross API support seems nice. Gradually integrate vulkan in old OpenGL codebase could be a thing with that.

[Azpegath]

"a rigorous memory model"
Uh, wasn't the whole point of these shiny new APIs the fact that they wanted to get rid of this?

Like Zan Lynx said: They wan't to clear up their specification of the memory model, so it is consistent and developers know what to expect and when.

[duby229]

in linux amd crossfire don't work (even in windows it sucks) and sli is become less and less necessary, having a single and powerfull gpu is much better, only 1% of gamers use it

Crossfire and SLI is something different called homogeneous multiGPU . They send every other frame to every other GPU The way DX12 and Vulkan do it is totally different, more compute-like called heterogeneous multiGPU .

https://blogs.msdn.microsoft.com/dir...to-the-future/
Heres an article about DX12, but the situation with Vulkan will be very similar.

[SaucyJack]

in linux amd crossfire don't work (even in windows it sucks) and sli is become less and less necessary, having a single and powerfull gpu is much better, only 1% of gamers use it

Explicit multi adapter isn't crossfire or SLI. Everyone would be using it if it was available. You could pair your old 290 with a new 480 for example. Or even a 290 with a 1080. On the few benchmarks performed on it, it scaled very well even on grossly different hardware (radeon+nvidia).

[tegs]

They need to slow down. What is the point of "Vulkan Next" if adoption of the original Vulkan is pretty much non-existent? Maybe they should spend this effort into improving development tools and aid developers in implementing it. I know having support in Visual Studio would go a long way for many devs.