The goal of Vulkan is very similar to that of Apple’s Metal API for iOS and OS X. Vulkan is a low-overhead graphics API that gives developers more direct access to the graphics processor and control over how their graphics are rendered.
In today’s announcement, Google argues that current APIs are “not designed for multi-threaded use, requiring synchronization with locks around calls that could be more efficiently done in parallel.”
The Vulkan API is being developed and maintained by the Khronos Group — the organization behind OpenGL,WebGL and similar standards. Indeed, Vulkan is essentially the next generation of OpenGL. The Khronos group first announced Vulkan in March, together with a range of software and hardware partners. But the API needs support from the operating system, too, and at the time, Google wasn’t yet on board.
You can partially credit AMD for this becoming reality as well, as Vulkan was modeled after their Mantle API.
It looks like a highly competitive race between Apple’s Metal, Khronos Group’s Vulkan, and Microsoft’s DirectX 12. This alone should ensure that some nice performance improvements are on their way.
I don’t see anything competitive or any race. Apple and MS will promote their APIs for their platforms and Vulkan will be the same unwanted children as OpenGL has always been. So you may see Vulkan used in dirt cheap and Valve games with the exception of a few jewels and that’s it, MS will fund and support any big studio making a new game.
ms wont fund nearly as much as you think they do. they couldn’t afford to. yes ms will promote dx12 with win 10 and already are, but dx12, metal, and vulcan are all based on mantle so it should make adoption easier not harder. linux does not do dx12 and steam machines use linux so i think more pc games will release opengl versions. steam sells more games than microsoft by far and they are pushing their own linyx os so they can control the store and get the money. valve has to get behind vulcan when before steam was windows only. google/chrome is also linux. so thats two big supporters with deep pockets that weren’t there before
Umm… Isn’t Vulkan on android kind of a bad idea? It’s already hard enough to develop for. And then you have the phones that will never, ever be updated, the bad drivers, the generally small team sizes for mobile productions, the plethora of strange hardware configurations…
No, they’re not. Steam on Linux is in a really bad shape and hasn’t moved much for a while now. +90% of current games are DX only and I’d say that +99% of all games are DX only. I’m not talking about Steam only, just for clarification.
On my very own case, I have 75 games on my Steam library and only 25 of them work on Linux. But about 10 of them are spins from the Half Life franchise so lets say 15.
So you bought both The Witcher 1 & 2 games? Too bad you will have to go back to Windows to play the first one.
So you bought both Borderlands 1 & 2 games? Too bad you will have to go back to Windows to play the first one.
So you bought the Bioshock franchise pack? Too bad you will have to go back to Windows to play the two first parts.
Adreno, Mali, Tegra and PowerVR? Just one more than PC if Intel sticks to PowerVR.
yes esp the vast majority of games are still windows only on steam. but compare that to two years ago when every games on steam and stream its self was windows only. remember when every pc was imp or ibp compatible? when every home processor was 386, 486, etc…aka intel? you have to start somewhere and grow. and companies can make comebacks. look at apple. from a real competitor, to nearly out of business, and back to relevant. microsoft has an advantage that they can issue ultimatums with dx. with opengl you have to spend the time herding everyone into the same direction. dx 12 is already out. kronos need to get vulcan out and deployed. with everything being based on mantle vulkan is kronos’s chance to catch up. kronos has to pull an apple. vulkan can be their ipod.
once opengl and d3d were pretty much equal. then dx pulled ahead because you just needed a decision not a consensuses, ms could roll up their sleeves and get to the real work faster, more working less debating. vulkan is a chance to start from equality again, a new starting line, a new race rather than being so far behind in the old race. nvidia had an edge in dx11. look at dx12. an amd 7970 beat a 980ti when it sells for about 1/2 the price. http://www.neogaf.com/forum/showthread.php?t=1098789 mantel based apis reset the game.
opengl is already there and beaten handily by dx. vulkan is not already there and can be the equal of dx. apple was already there and nobody cared. then ipod changed the game. when is the last time you went a day without hearing the word iphone. you still dont hear the word macintosh. just because nobody cares about your old tech does not mean you cant have success with new tech.
DX12 and Vulkan are almost the same thing. Both APIs being equal will maintain the status quo. Microsoft in their Windows + Xbox strongholds and Khronos mending for some attention in dirt cheap Android games, those blue moon games that happen to launch for Linux and the re-re-re-release of the good old Valve games.
A lot of game engines nowadays have full support for OpenGL as well as Linux and the vast majority of 3D programs use OpenGL for their viewport code.
That’s not to say that OpenGL 4 has some major issues under the hood, but that is precisely why it’s being replaced by Vulkan. Plus if the open standard for graphics disappeared, then the entire concept of FOSS will take a nasty hit as it may screw Linux beyond repair and lead to many open projects having to go back to the days where application developers had to roll their own graphics code.
No, they’re not. Steam on Linux is in a really bad shape and hasn’t moved much for a while now. +90% of current games are DX only and I’d say that +99% of all games are DX only. I’m not talking about Steam only, just for clarification.
On my very own case, I have 75 games on my Steam library and only 25 of them work on Linux. But about 10 of them are spins from the Half Life franchise so lets say 15.
So you bought both The Witcher 1 & 2 games? Too bad you will have to go back to Windows to play the first one.
So you bought both Borderlands 1 & 2 games? Too bad you will have to go back to Windows to play the first one.
So you bought the Bioshock franchise pack? Too bad you will have to go back to Windows to play the two first parts.
That’s a little bit like saying electric cars don’t have a future, because there are so few of them as compared to older gasoline-driven cars.
There’s obvoiously a trend towards Linux in terms of games, and Steam’s CEO has given some very good reasons for it.
Steam OS will be interesting, not just because of the impact on the gaming community, but because it’s going to be such a graphics-oriented systemm – possibly the first geared mainly towards graphics since SGI’s (now defunct) IRIX.
That means all unnecessary overhead will be stripped away, and without that extra baggage, Steam OS may well be of interest for #D designers as well, since gaming and 3D design make similar demand on the system. So I wouldnæt be surprised is Steam OS became the go-to operating system in the 3D industry at some point.
Apple swiftly developed and implemented Metal prior to Vulkan was announced, much less finished, in a large part because development of OpenGL has been rather glacial over the years. They couldn’t wait for the committee design would get their act together, so they moved ahead. Additionally, Apple itself is a member of the Khronos group, so they are helping the effort along there too.
As for Blender, it’s not even on OpenGL 4 yet, which is why things like OpenSubdiv don’t work properly on OS X. Although its hopelessly behind in this area, Blender could potentially see some rather substantial performance gains if it were to move to any of these Mantle-llke API’s such as Vulkan, Metal or DX12. Although it may seem silly to adopt Metal or DX12, many game engines (say, Unity) happily support multiple graphics APIs
And yet most games developed with such engines don’t have an OpenGL version. Same with the vast majority of 3D programs anchored in ancient versions of OpenGL.
Blender, I’m looking at you.
But your analogy fits so perfectly. If you buy an electric car now you will have all the early adopter disadvantages. No recharging stations near your home or most of the places you want to go, no official service in hundreds of kilometers, endless delays getting spare parts and the feeling that if you waited just a couple more years you could have gotten a much better EC and probably cheaper.
The sensible thing is to wait until they’re more widespread. Same as getting into Blender, it got popular when it lost a great deal of its amateurish and hellish aura.
Except a good chunk of the immediate mode code has been replaced for version 2.76 (which means some modes are now a lot faster). Now it is true that some areas have not seen speedups yet with no VBO use currently, but that will likely be resolved by 2.8.
Sure, there’s the early adopter thing, but I wouldn’t be surprised if Nvidia and AMD move quickly to support it in their drivers.
No, looking forward, it’s good to have it on there. You won’t be getting the kind of low driver overhead you’d get from Metal from GLES2/GLES3. This isn’t necessarily something for everyone to use, but Unity Tech or EPIC or the larger studios do have the resources to support another renderer backend. Plus, at least in theory Vulkan drivers have less surface area for driver vendors to mess up and much better options for validation on the developer’s side. Old phones not getting updated isn’t such a big deal, mobile moves really fast and the most interesting customers are already on a 2-year phone replacement plan anyway.
It’s not the number of vendors, it’s the generally bad and wildly varying driver quality that is the problem. It’s impossible to test this without a zoo of devices. On PC at least you can reasonably ask your customers to update their (already “good”) drivers, on Android that’s not really possible.
My point is that OpenGL is already there and no one is paying it any attention.
There’s no point in supporting OpenGL for Windows when you’re already shipping DX anyway, the drivers are just that worse compared to DX. The “cross-platform” aspect is irrelevant when you end up spending those “code savings” in debugging ten times over. Linux or Mac ports tend to happen after-the-fact, for a more limited set of target devices, so you don’t need to maintain an OpenGL backend for the entire development time. For Vulkan, things might change, because DX12 only works on Windows 10 and maybe (just maybe) drivers aren’t going to suck.
First of all, even if somehow all non-proprietary graphics APIs died, FOSS developers could still use those proprietary APIs anyway, just like they can (and maybe should) use DirectX on Windows or Metal on Mac OS. I don’t know what “time” you are talking about when you’re talking about “rolling your own graphics code” (doesn’t everybody have to do that still?), but if you’re talking about the time before proprietary graphics hardware, when people wrote software renderers: Those were the best of times and if it wasn’t for speed advantage of those proprietary chips they could all go die in a fire and we’d be living in the wonderful world of software rendering ever after. Briefly, it seemed like Larrabee might’ve made that a reality, but it just wasn’t quite fast enough to survive in the (consumer GPU) market.
