A few weeks ago, NVIDIA and Epic announced the inclusion in the Unreal Engine of support for the technologies implemented in the Santa Clara RTX 2000 and RTX 3000 cards. This which may seem somewhat minor is a knock-on effect that can affect the graphics card market.
The Unreal Engine is the most widely used graphics engine in the videogame industry, the addition of NVIDIA’s proprietary libraries in the main branch of said engine can put AMD in a bad situation.
NVIDIA at the forefront
It is clear that although AMD is also a participant in the graphics card market, the reality is that the most important graphics advances that have occurred in recent years occurred in NVIDIA’s Turing architecture, launched as the RTX 2000 range between 2018 and 2019.
Technologies such as Mesh Shaders, Variable Rate Shading, the addition of specialized hardware to accelerate algorithms for artificial intelligence such as Tensor Cores and hardware to accelerate Ray Tracing in the form of RT Cores were seen for the first time in the Turing architecture. .
But any new technology in terms of graphics hardware needs two stages for adoption, the first is its integration into the graphics APIs and the second into the graphics engines, which are used by many games and which allow them to adopt these technologies directly. to any game that makes use of this engine, which in turn accelerates the adoption of the hardware.
The NVIDIA-Epic pact, a chess move by NVIDIA
A few weeks ago, NVIDIA announced full support for Fortnite for NVIDIA card ray tracing and its real-time super-resolution technology called DLSS. This, which in principle is something that may seem minor, is much larger than what is seen with the naked eye, the reason is that Fortnite is the game with which Epic deploys the improvements of its Unreal Engine and any improvement over said game translates into an improvement that the rest of the games that make use of said engine can take advantage of.
All this has resulted in a series of improvements that any game in development or that has been developed under the Unreal Engine 4 can use, and all this focused on NVIDIA hardware. A chess move against its direct rival, AMD, which can have an impact on the end user’s purchase decision.
The fact that the new libraries are included as standard in the Epic graphics engine means that game development studios that make use of the Unreal Engine 4 can apply them in their games, both in development and retroactively and since the update is focused exclusively in NVIDIA the consequences can be disastrous for AMD that does not have the way to counteract them, especially those that make use of hardware not yet available in the Radeon such as Tensor Cores.
The key is native NVIDIA DLSS support on Unreal Engine
Both DLSS and DLSS 2.0 are algorithms based on artificial intelligence that take advantage of the high calculation power of the Tensor Cores to generate a higher resolution version of the internally rendered image, causing that it is not necessary to have all the power to reach certain resolutions and making the frame rate much higher.
On the NVIDIA page we can read:
The NVIDIA DLSS is a deep learning neural network that improves frame rates and produces crisp, beautiful images for your games. Delivers the power needed to maximize Ray Tracing performance and increases output resolution. It is available for the first time in the main branch of the Unreal Engine as of version 4.26.
Outside of the propaganda, if we take into account the number of titles under the Unreal Engine 4 and the fact that for now only NVIDIA has hardware like the Tensor Cores, then it is very clear what the Santa Clara strategy is.
The DLSS only requires that depending on the resolution and the GPU, the game leaves a few milliseconds, which are in the table, for the GPU Tensor Cores to apply the DLSS algorithm and automatically generate an image with 4 times more pixels, but without the computational load or the time required to render the image at native resolution.
The adoption of the Unreal Engine in the video game industry
One thing that is very clear is that one of the most difficult things when developing a game is the development of a new graphics engine that makes use of the newly released technologies, since this is a huge expense not only in time and resources before being able to make a new game that takes advantage of new technologies but also in the fact of internal training within the studies.
The fact of supporting new technologies in Unreal Engine 4, despite being an already veteran engine, means not hitting the reset button in the know-how and not having to touch the fundamental structure not only of many games but of methodologies learned in the different studies, which facilitates the adoption of new technologies in a less expensive, fast and easy way.
Not all studios can afford the development of a high-caliber graphics engine and the launch of a game, in recent years we have seen how Electronic Arts has choked on its Frostbite, we cannot forget the case of Cyberpunk 2077 where development of a new engine at the same time that the game was made has given very big problems to the final development of the game.
If we combine the departure of the new generation consoles and the adoption of certain technologies on the PC, it is clear that we are facing a technological race in which whoever first takes advantage of the new specifications will have a knock-on effect on the competition. That is why there is no time for new graphics engines from scratch and outside companies like Ubi Soft that with their Anvil Engine have reached a certain level of maturity, the rest will be forced to use the Epic engine.
In any case, it must be taken into account that the fact that Epic has added a series of NVIDIA technologies in its Unreal Engine does not mean a massive adoption by the different development studios, especially if we take into account the factor of consoles that are based on RDNA 2.
But the PC market is a separate market, no one who has spent money on a next-generation Gaming PC is going to settle for converting a console version, and although many studios will be tempted to do so, we cannot forget that the PC on a technical level has always been at the forefront, especially in graphics.
With its RDNA 2-based AMD RX 6000 AMD is going to try to rob NVIDIA of some of its huge market share when it comes to graphics cards, which is over 70%. This difference is crucial when deciding whether to support certain technologies or not, if NVIDIA were in the position of AMD in terms of market share it would not have the level of influence that it has.
And let’s not forget that it is not only the Unreal Engine 4, the DirectX 12 Ultimate is nothing more than the adoption of the technologies that NVIDIA added in Turing within the most used API, for the moment Microsoft in order not to leave AMD at a disadvantage on PC it has not included elements equivalent to DLSS or noise elimination for Ray Tracing as minimum specifications in order to give AMD room.
AMD is out of time with NVIDIA
The reasons why NVIDIA has been able to convince Epic to integrate its proprietary libraries into its graphics engine are mainly two:
- Two years of advantage regarding the implementation of certain technologies translates into two years of advantage of tool and library development.
- AMD decided to throw away non-proprietary libraries and tools, it does not make any effort to develop proprietary libraries to ensure that such technologies are adopted for AMD cards. So there is not as close contact as NVIDIA has with developers.
All this does not mean that AMD graphics cards are bad or that we have gone crazy in an ode to NVIDIA, simply the fact that the most used graphics engine in game development is already optimized for the use of almost 75% of the graphics cards that are being sold, which are of a specific brand, make AMD have a very uphill job in order to face NVIDIA.
At the moment and in the current situation, AMD cards hold the type with dignity and are a more than valid option, it is clear that the objective of NVIDIA is the adoption of technologies that give them an advantage over AMD. The fact that the studios do not have to implement them as they are already in the most used graphics engine together with the market share leaves the ball in the hands of the developers in terms of titles in development and with it the useful life of the cards graphics.
The next few months of 2021 will be very important, as then we will know if NVIDIA’s strategy of forcing development in some sense has been successful or not.