Mesa Vulkan Branch Published For Intel Linux Support

[phoronix]

Phoronix: Mesa Vulkan Branch Published For Intel Linux Support

Following this morning's long-awaited Vulkan 1.0 release with our lengthy write-up and initial thoughts, Intel has joined NVIDIA in offering a Linux driver for Vulkan...

http://www.phoronix.com/scan.php?page=news_item&px=Intel-Mesa-Vulkan-Published

[chuckula]

Thanks to Intel & LunarG for getting the first open source Vulkan driver available on launch day.

I think it would be rather classy of a few of the usual suspects who constantly spew hatred at Intel to acknowledge the positive aspects of their open source support, especially when a certain three-letter company who gets all kinds of praise around here has decided that Vulkan on Linux isn't really worthy of their support.

[duby229]

Thanks to Intel & LunarG for getting the first open source Vulkan driver available on launch day.

I think it would be rather classy of a few of the usual suspects who constantly spew hatred at Intel to acknowledge the positive aspects of their open source support, especially when a certain three-letter company who gets all kinds of praise around here has decided that Vulkan on Linux isn't really worthy of their support.

That's snarky and unfair, but I certainly do agree with that AMD should have had Vulkan running on amdgpu today. I don't think AMD realizes that bad press is worse than no press. Even if todays launch only had a footnote about AMD support, it would have been better than this outcome.

EDIT: I seriously think AMD needs to put some very smart people in charge of planning press manipulation. They had plenty of time to prepare for todays launch and they should have been ready to put the best possible light on this situation.

[gQuigs]

I'm really happy to see it posted as part of mesa. That makes it much easier to get it the same why I get mesa releases today (or will). I really hope AMD releases it as part of mesa too.... It seems counter productive not to..

I'm also curious if one Gallium driver gets it if others will be have a significantly easier time to get it as well.

[boffo]

Thanks to Valve & LunarG for getting the first open source Vulkan driver available on launch day.

Fixed, just to make you more angry XD.
I don't know what is wrong with people praising Amd, but on the other hand to my knowledge no one is complaining with Intel support.
Anyway I wish Amd would opensource Vulkan in alpha/beta sooner, to attract the interest of new developers.

[boffo]

I'm really happy to see it posted as part of mesa. That makes it much easier to get it the same why I get mesa releases today (or will). I really hope AMD releases it as part of mesa too.... It seems counter productive not to..

I'm also curious if one Gallium driver gets it if others will be have a significantly easier time to get it as well.

That would be nice, it would be nice to have as much shared code as possible, but in the article it's stated that there is not so much.

[log0]

Fixed, just to make you more angry XD.
I don't know what is wrong with people praising Amd, but on the other hand to my knowledge no one is complaining with Intel support.
Anyway I wish Amd would opensource Vulkan in alpha/beta sooner, to attract the interest of new developers.

The driver mentioned is from Intel, has nothing to do with Valve or LunarG.

[zanny]

The timing on this is pretty bad. This cannot make the 11.2 merge window, so Vulkan in Mesa cannot land at the earliest until 11.3, which means it won't be in Ubuntu 16.04, which means we are four years out from the first release of the most popular distro that most people install actually supporting Vulkan.

The consequences of this are compounding. SteamOS usage is disastrously bad - it had literally no effect on Steam on Linux usage stats because the launch was botched and terrible. Any developer targeting Linux today is targeting Ubuntu, and when you target Ubuntu you target the latest LTS. We already know developers hate writing OpenGL code, but the new question is why in the world would you write your game using Vulkan right now?

DirectX 12 has been out for over half a year, and it is... DirectX. If most PC developers could they would put a ring on the name and marry it, because nobody ever uses OpenGL on Windows. The natural state of affairs is to keep doing what you have been doing - writing platform locked DirectX games, now using DX12, because these people have never touched a Khronos product before and the heads that steer development will never make change for changes sake. Even then, DX12 is Windows 10 only and market adoption is way too low to ever actually use it.

However, Vulkan support on Windows is looking... ok. You can't actually use it, because all the vendors have dropped the ball devastatingly by only supporting extremely new GPUs with their drivers. Intel doesn't even officially support Haswell, where new parts were shipping last year. Nvidia lied and is only supporting Kepler, and AMD on Windows for some reason actually has the best Vulkan product support - all GCN GPUs are supported. Though that doesn't help that plenty of people still have 5xxx and 6xxx series hardware that is... unsupported, but could still run many modern games at fairly good framerates. The last AMD product using the VLIW5 model was the Trinity APUs released in 2012 - four years ago! And they won't get Vulkan support.

And then you hear about how open the standard is and how cross platform it is going to be. Linux on the Nvidia and Intel side are looking to be just as good / bad as the Windows side, but then AMD drops the ball with no driver, and only promises a GCN 1.1+ driver... eventually.

And Android is the worst of all the platforms targeted, because this means Vulkan will only be supported in Android 6.1 or Android 7 - a version that has not even been released yet, in an era where Android 6, a year after release, still only has 1.4% market share, and 5.0 - released two years ago - is only 1/3 the market. 2/3 of the market of Android devices is on some version of Android 4, which is up to five years old now. Again, you cannot use Vulkan here today, because the first version that could even potentially mandate Vulkan support is the one that isn't even out yet! And the history of Android releases shows it can be upwards of 6 years until the majority of phones are on that version or later.

So put yourself in the shoes of a developer right now. Vulkan on Windows doesn't support 2/3 of GPUs out there. Vulkan on Linux supports even less. Vulkan on Android will take probably half a decade to land on a majority of Android devices. What API are you using? Not Vulkan - or DX12 - you are just going to keep using what you know, DX11, because none of these APIs are going to be viable targets for new releases for at least a couple years, or until Intel / Nvidia / AMD backport Vulkan to every GPU from the last half decade at the least.

Until then, Vulkan can only be a secondary render target - extra work on top of your still necessary DX11 game renderer, plus the OpenGL ES one if you want to target Android (or OpenGL if you drop Android yet target Linux / OSX).

And then the final nail is that even after all these platforms, iOS remains a huge segment of gaming revenue that without Vulkan support, if the Metal Vulkan compiler does not work on iOS, the maintenance of multiple renderers still.

I'm not saying this is the end of the world, but this does mean we cannot expect mainstream Linux gaming to be a thing until at least Vulkan is actually usable, because developers have demonstrated their refusal and the driver developers inability to provide a working OpenGL environment that is conducive to gamers, and it looks like you cannot reasonably target Vulkan except as an extra API to target for at least five years, and most developers don't have the time or money to support so many different graphics APIs at the same time.

[boffo]

The driver mentioned is from Intel, has nothing to do with Valve or LunarG.

So what happened the LunarG driver?

[Pecisk]

So what happened the LunarG driver?

It was used as prototype/test bed for ideas to be fleshed out. That was it's goal. LunarG also used it to develop SDK, while this driver got rewritten from scratch under Intel supervision.