Hi,

Any thoughts on how the coming DirectX 12 outroll will impact the industry in general and also certain segments of the market?

For those who do not know, it comes with promise to help developers have more performance advantage in low level coding and such issues by providing a thinner API layer between developer and hardware. Subsequently, game development costs in these related areas should be leaner than in the past while increasing access to optimizations such as threading.

In the near term the effect is approximately nothing.

It is not released. The official stance is that it is tied to the Windows SDK, and the next update is Windows 9.

Unless Microsoft decides to change their prior stance and push out an update to Windows 7, or decide to break out back out from the Windows SDK and back into a separate library, it will have very little effect on everyday games until something happens that drives people to migrate from Windows 7 to Windows 9.

Also the focus of the industry is blurred right now.

You will have many more people having high level meetings about what devices to support, will they spend resources on a steambox version, how many Android versions will they put out, that sort of thing. DX12 will be mostly ignored until it starts causing problems.

DX12 is not something I personally will be looking at, couldn't care less about it.

But then I have refused to use windows 8, so I am not the best litmus test

i'm focussed on windows development, and i couldn't care less RIGHT NOW about dx12. because, it's 1.5 years away for general release. so essentially, it's a "we'll see" for me.

Depends.

If it's available on Windows 7 - happy days.

If it has feature levels allowing it to run on downlevel hardware - happy days.

If Microsoft hit those two sweet spots with it, allowing it to be usable by most people on day 1, then it will be successful. Remember: this is what they did with DirectX11 (it ran on Vista and had feature levels) so there is precedent. Hopefully they're aware that if they want this to be successful they'll have to modify some of their current attitudes.

Win7 support is critical; while it might be sometime away once we get there Win7 is likely to have a huge market share still so not supporting it would be a problem.

Really what it needs to be is DX11 features with a new front end; so even if Win7 only get DX11 features, Win8 the 11.1/2 and Win9 more still at least the common API set will be there.

What I'm hoping for most of all is that it'll prove the concept will work, there are good games and it finally forces the OpenGL ARB to get their head out of the instance-all-the-things sand and embrace this way of working.

DX12 doesn't seem to introduce new features, just a leaner API. A leaner API optimizes performance (primarily with respect to CPU usage) but it's also a pain in the ass to code for. Look at the Mantle slides for a glimpse of that. Plus there's a risk that MS will screw us on OS support, as phantom mentioned. My take is that it will be an advertising point for AAA games but have very little impact on the industry otherwise.

it finally forces the OpenGL ARB to get their head out of the instance-all-the-things sand and embrace this way of working.

Based on the latest stuff out of the GL camp, they're staying the course. Their approach appears to be focused around GPU driven rendering (MultiDrawIndirect) and bindless/sparse arrays. That plus a few other toys like persistent mapping/fences, buffer storage, and gl_DrawID. The slides actually admit that DrawID doesn't work properly everywhere.

Yeah, that's why I'm hoping something like this might shake them up a bit; if Mantle and DX12 can do enough that OpenGL suddenly becomes the bottleneck maybe, just maybe, it'll force something to happen.

Those GL slides are just amusingly because right now it's not so much 'approaching zero driver overhead' as 'NV best practises' until AMD get up to speed (no one really expects Intel to work correctly after all), which in a way makes their persistence on this all the more disappointing as it's only half the solution and not even half everyone can do anyway :|

With the "Approaching Zero Driver Overhead" talk I thought they'd make an announcement along the lines of "And now all these extensions will be supported in Intel, nVidia and AMD!" but it didn't happened lol