AMD’s graphics cards have historically provided slightly less efficient performance than Nvidia’s GPU solutions, especially in games that are built on Nvidia’s proprietary technology, GameWorks.
As a result, Nvidia graphics cards have enjoy a better power-to-performance ratio than AMD cards in DirectX 11 games.
However, things are changing with the proliferation of DirectX 12 and its competitor, Vulkan.
Vulkan and DirectX 12
An API (Application Programming Interface) is a set of tools that allows developers to easily build software and applications according to a uniform set of rules. The interface essentially provides the ‘building blocks’ for developers to create their software.
Vulkan and DirectX 12 are both low-level APIs geared at modern gaming needs. Vulkan is an open-source API developed by Kronos and based on AMD’s Mantle code, while DirectX 12 is a closed API owned by Microsoft and only compatible with Windows.
Although Vulkan is still fairly new, it is quickly demonstrating the advantages of open-source development, as the API is compatible with a great amount of platforms – such as Android, PlayStation 4, Xbox One, Linux, and Windows.
This makes porting games between platforms much easier and minimizes performance loss.
Vulkan has only been implemented in a few games so far, one of which is Bethesda’s DOOM reboot.
Vulkan has recently been implemented in mainstream games, but it is already making a significant difference in optimisation.
Whereas AMD graphics cards were previously limited in the OpenGL-powered DOOM, activating Vulkan allows them to reach their full potential – increasing performance by up to 66% on the Radeon Fury X.
Although Nvidia cards also see a performance increase using the Vulkan API, it is not nearly as significant – with some older cards showing no difference at all.
A set of benchmarks published by ComputerBase.de outline the performance difference offered by Vulkan, showing that AMD Radeon graphics cards enjoy a massive boost to performance, and are set to continue to use Vulkan to implement their asynchronous compute effectively.
Check out the benchmarks below: