GDC 15: Khronos Acknowledges Mantle's Start of Vulkan

Subject: General Tech, Graphics Cards, Shows and Expos | March 3, 2015 - 03:37 PM |
Tagged: vulkan, Mantle, Khronos, glnext, gdc 15, GDC, amd

View Full Size

Neil Trevett, the current president of Khronos Group and a vice president at NVIDIA, made an on-the-record statement to acknowledge the start of the Vulkan API. The quote came to me via Ryan, but I think it is a copy-paste of an email, so it should be verbatim.

Many companies have made great contributions to Vulkan, including AMD who contributed Mantle. Being able to start with the Mantle design definitely helped us get rolling quickly – but there has been a lot of design iteration, not the least making sure that Vulkan can run across many different GPU architectures. Vulkan is definitely a working group design now.

So in short, the Vulkan API was definitely started with Mantle and grew from there as more stakeholders added their opinion. Vulkan is obviously different than Mantle in significant ways now, such as its use of SPIR-V for its shading language (rather than HLSL). To see a bit more information, check out our article on the announcement.

Update: AMD has released a statement independently, but related to Mantle's role in Vulkan


March 3, 2015 | 04:13 PM - Posted by Anonymous (not verified)

So mantle came up to the surface and formed Vulkan, with the SPIR-V(intermediate language) implementation.

"SPIR-V unifies graphics and computation. Both OpenCL 2.1, also announced today, and Vulkan will ingest programs in SPIR-V intermediate code."*

So AMD/HSA foundations HSA-IL, will it maybe become supplanted buy SPIR-V IL or will there be some form of HSA-IL conversion layer that will allow this, and other shader code/GPGPU code to be converted/recompiled and passed on the Vulkan VM/front end. So this is definitely good news, I'm very interested in seeing what the HSA foundation will have to say about this, and Neil Trevett's( current president of Khronos Group and a vice president at NVIDIA) comments, and Vulkan's mantle based/influenced design suggests, that there has been some indirect cross pollination of Ideas between the HSA foundation, and the Khronos group, as many of the companies involved are members of both organizations.

* ArsTechnica

March 3, 2015 | 11:54 PM - Posted by Patrick Proctor (not verified)

I think Spir-V is going to supplant much HSA-IL. Even Intel is fully converting its GPU offerings over to it, and those GPUs are going on every single desktop and server chip from Skylake on in the name of compute density and to fight off Zen with everything the blue dragon can bring to bear.

It'll be interesting to see if AMD and Intel can eat into Nvidia's accelerator marketshare in 2016. Having CPU/APU chips which synergize extremely well with their own accelerators puts AMD and Intel in a much more favorable position, especially since Nvidia has no enterprise moves it can make promptly.

Knight's Landing(card and chip), Skylake EP/EX, OpenMP, OpenCL 2.x
vs.
Zen, Fiji FirePro, HSA, OpenCL 2.x
vs. Kepler 2.0, Pascal*

*may come so late into 2016 to be able to avoid the damage.

March 3, 2015 | 04:26 PM - Posted by ZoA (not verified)

While initially I was quite disappointed with AMD decision to quit public SDK and make mantle API SDK exclusive only for it's partners, the more I think about it the more it looks as best thing to do for AMD and for us consumers.

Mantle as AMD developed and maintained API would have hard time gaining support of Intel, Nvida, ARM and other competitors for political reasons. But AMD handing Mantle code to independent party such as Khronos for modification, development and maintenance under different neutral “Vulkan” brand allows that code to benefit much wider consumer base without threatening anyones prestige.

On another hand it benefits AMD too because they don't have to maintain and provide support for open SDK themselves as that will be done for by Khronos group. This will allow them to focus their limited resources and manpower on other projects, including continued development of Mantle as AMD partners only API.

My only remaining beef with this AMD decision is that it gives time for DX12 to entrench itself as dominant API before Vulkan is fully finished. Mantle with it's significant library of published titles, almost two years of time advantage over DX12, and almost hundred beta participants had capability to undermine that dominance, and provide alternative for us who hate spending money on M$ products. I think Vulkan will have much harder time doing that, especially as Vulkan development is going significantly slower then DX12. They have jet to produce even a public demo of Vulkan working (unless you consider Mantle itself as sort of Vulkan early development demo)

March 3, 2015 | 05:12 PM - Posted by Anonymous (not verified)

You are 100% correct in your last point. AMD should use the leverage afforded to them by their head start with Mantle to quickly migrate their partners to Vulkan.

I'm sure AMD will keep Mantle as a tag along for experimental and PR features, though, and that makes me pretty sad.

That being said credit must be given where it is due: much kudos to AMD, and well-deserved, for following through on their promise of allowing industry access to and free use of Mantle. I can honestly say I am glad I was wrong about AMD's intentions.

March 3, 2015 | 05:50 PM - Posted by Bill D Strong (not verified)

First, there is a video demo from Imagination Technologies show a prerelease driver with a demo ported from OpenGL ES 3.0.

Second, Once the spec is finished, AMD will probably have the easiest time of implementing the spec. you can consider Mantle a 0.1.0 alpha of the spec. It looks like any of the games that are already shipping Mantle code should be able to port over to Vulkan quickly, the biggest area will be porting their HLSL shaders to SPIR-V.

This should allow at least one vender to have an implementation with games shipping within the year, or maybe on the day of the final spec.

This is good news.

March 3, 2015 | 04:53 PM - Posted by MaestroGK (not verified)

There's also some Vulkan source code:

vkCmdBindDescriptorSet(cmdBuffer, VK_PIPELINE_BIND_POINT_GRAPHICS,
textureDescriptorSet[0], 0);
vkQueueSubmit(graphicsQueue, 1, &cmdBuffer, 0, 0, fence);
vkMapMemory(staticUniformBufferMemory, 0, (void **)&data);
// ...
vkUnmapMemory(staticUniformBufferMemory);

Now let's check some Mantle64.dll code:

grCmdBindDescriptorSet(cmdBuffer, VK_PIPELINE_BIND_POINT_GRAPHICS,
textureDescriptorSet[0], 0);
grQueueSubmit(graphicsQueue, 1, &cmdBuffer, 0, 0, fence);
grMapMemory(staticUniformBufferMemory, 0, (void **)&data);
// ...
grUnmapMemory(staticUniformBufferMemory);

Vulkan is 100% similar to Mantle

March 4, 2015 | 10:59 AM - Posted by Goofus Maximus (not verified)

With this announcement, and AMD's announcement of changing their focus on Mantle, is it possible to say that...

"AMD has passed the Mantle of development to the Vulkanologists of Khronos"?

Is Vulkan just "Mantle 2.0"?

March 4, 2015 | 11:54 AM - Posted by Anonymous (not verified)

Mantle is now AMD's internal/partner new technology development API, and any new improvements in Mantle will find their way into the Vulkan ecosystem. Vulkan will be the public open face of not only Mantle, but the other industry players in the gaming market, Khronos being the industry open standards group for the various parties who make up the Khronos group's membership and standards committees.

So this(Vulkin et al) is the de facto public version of not only Mantle, but also the other makers API inputs. Mantle will be retained internally for AMDs development use, and to share with AMDs game development partners, for sure this will allow for much cross pollination of Ideas among the various market members, both proprietary and open standards based.

The mobile and Steam OS, Linux/other markets, as well as others that utilize the Khronos group's open standards will continue to get a look at AMD's internal Mantle SDK going forward, Khronos, and its various standards, will be the open standards(derived from Mantle/other) public face for a good part of the devices/gaming industry Graphics API/middleware/GPGPU open standards ecosystem, it always has been, and will continue to be.

March 4, 2015 | 02:42 PM - Posted by Anonymous (not verified)

And will nvidia's pr guy known as Ryan present his public excuses for the pile of sh$t he wrote yesterday ?

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.