AMD’s Mantle API has already proven that it is optimum for the players to give them a boost in gaming and also in rendering processes. Microsoft’s DirectX 12 specifications are also underlying for the optimization as an important feature, previously we have seen in Demo that it’s performance ratio improved to 70% in comparison with DirectX 11. Now, from the exposure of the latest news we come to know that DX12, although similar in the rendering process with DX11, but each thread rendering time reduced from 9ms to 4ms.
DirectX 12 performance Leaked: Rendering time Twice faster in Comparison with DX11
From the view of the above picture, D3D11 rendering process to go through App Logic (application logic layer), D3D11 (D3D runtime), UMD (user model-driven), DXGK (DXGK layer), KMD (kernel mode driver) until the completion it goes till Present. Each stage must consume a little time. And if you see the above bar at thread 0, UMD consumes most of the time after KMD.
Now in the other slide of DirectX 12, most of the rendering process has been gone, but D3D runtime, UMD significantly reduced with time, even without the KMD process (it may be rendered with too little cannot see), greatly reducing rendering time. DirectX 12 splits the load across multiple cores while DirectX 11 does everything on a single core.
DX11 take more than 9ms in order to complete the whole process, while the DX12 with only 4ms.
Of course, this does not mean that double the rendering speed will be able to give the double the performance, after all, there are bottlenecks, but DX12 optimization of the underlying hardware is indeed very beneficial to developers to enhance efficiency, improve performance.