The thing that is interesting about Vulkan is that it's a cross platform api, it will allow developers to not be tied down by Microsoft, it will allow better gaming on Mac, and other OS's like Steam OS for example. Future of PC gaming is looking a little more interesting that's for sure.
This is not a proper Vulkan implementation. The way Vulkan was implemented prevents it from "running at full speed"... Mantle was nothing but a "proof of concept". Vulkan and DX12 are the resulting products.
I don't think this is suppose to be seen as a first impression, like I said just a proof of concept. If it were, then it would be a horrible first impression. They aren't saying that it was released to be used, just that it can be done. Nvidia and AMD just gave limited resources to enable Vulkan support. CUDA over the years have increased performance significantly with optimization. I'm just excited to see Vulkan. And so far the reaction is positive to just having some Vulkan support!
In other news, Dolphin emulator sees an increase of up to 50% by switching to DX 12 (compared to DX 11). Source: PCGamer.com
Vulkan should see similar or greater speed increases than DirectX 12 and Mantle once proper drivers are released and games/benchmarking applications are optimised for it. My evidence of such a claim is that Vulkan is effectively Mantle 2.0 (look up the history if you don't believe this ), and DirectX 12 apparently borrows from Mantle also. Some games supporting Mantle may still only be Windows 10 if it means development is easier. In any case, a large number of games probably won't even support Mantle and those that do will probably be a little ways away yet. I think come 2017 it would be pretty sad if you're still running Windows 7 and expect full support to be handed to you on a platter . To put it into perspective, running Windows 7 is like using Android 1.5, or iOS 3.0. Microsoft supporting Windows 7 and 8.1 simply costs too much. There is no more income coming from these streams, and yet people still expect 100 percent full support for the latest hardware and support for gaming? There aren't too many companies supporting deprecated products so long. Moving gamers and home users to Windows 10 saves them money and lets them focus development on a properly built OS, not an OS that is patched to the hilt and is lucky there hasn't been any more issues with it. Companies still using Windows 7 and 8.1 are fine, because keeping security support for computers that run workstation apps is significantly easier than taking all other situations into consideration as well. I half suspect though that Vulkan will probably still work better on Windows 10 than Windows 7 or 8.1
Properly built? Have you actually used Windows 10? I've installed it on 2 laptops, 3 desktops, and 1 tablet. So far, it's the worst experience I've had with any OS. Based on the proven lack of stability, I can't find an argument for people to migrate to Win10. I've seen Win10 crash more times since it's launch than I have Vista, 7 and 8.x combined. As a customer, I expect MS to fully support a past, STABLE product at least until their current product is stable.
Aye, it shall live forver in Vulkan's veins. Like Quake code lives in Source Engine's heart. Don't listen to HeavyHemi's blasphemy.
So Valve's Source 2 engine will support Vulkan, with Valve licensing the source engine to other developers this is possible a lot of new games supporting Vulkan. I hope CryEngine and Unreal Engine jump on board.. or have they already?
Maybe i am being harsh, but if it's to early to judge then when can we judge? People are looking forward to it as it won't be limited to W10 but at this rate when it finally makes an appearance in games W10 will be an old OS. The idea of one api for all platforms sounds like less work for devs, but im not sure it is as it can't be easy to guarantee consistent performance across different operating systems, also sounds like a lot of extra work before and after release. Not even that long since a Dice dev claimed Mantle was held back by W7/8 while singing the praises of WDDM 2.0.
If the game engine is coded properly, it'll result in less work after release. As for the performance, it'll be quite tuneable.... The API is essentially the same for every OS. It allows "close to metal" programming, so the hardware can perform at peak.
Don't see how it could be less work, i doubt the api will not be tested on every supported OS. It's only close to, not direct to metal so don't see how there will not be differences in performance and stability depending on the OS used or the vendor driver used.
Unreal Engine 4 supports even android. And 1st demos we got were running very well on PowerVR while having exquisite graphics considering it was on mobile chip. Only thing which did not make me happy is that AMD/nV comparison. Not feature wise on HW, it matters little as it is partly driver's work to expose those. I saw that Vulkan actually allowed nVidia to have vendor specific extensions. That's one reason why I did not like OpenGL much. It increases work for programmer and chance of bugs. And it is likely one of reasons OGL was less popular than Direct3D. In the end I believe there should be 1 software specification and let each vendor's driver to deal with is. Each driver splits programmer's code based on HW needs. Why to do step back and split it partly in hands of programmer?
Having more alternative API's is another crucial step for windows-free gaming. To even have a choice in os other than windows for gaming on pc is a huge step forward that's taken time to achieve, but, thanks to companies like Steam, Nvidia, AMD and competition from Apple and Google, etc; this dream is quickly becoming realised.
Nope, Ti only has 1. AMD has 3, graphics queue or compute queue, a compute queue and a copy queue. Nvidia has 1 queue that does them all. AMD should be able to feed its pipeline faster, but whether or not that translates to real world increases is left to be seen. It's not just about how much data you can cram into the pipeline, its about how fast that data is crammed in and how fast it can be executed. I would imagine in the end it will benefit GCN cards pretty well though, especially in compute heavy engines/games.
not "async compute" bs once again. We can thank Oxide and AMD for misseducating consumer... sigh Spoiler https://forum.beyond3d.com/threads/dx12-performance-discussion-and-analysis-thread.57188/page-51
