GPUs Latest news Software

AMD talks about the future of RDNA and wants to make NPCs in games smarter with AI

In late 2022, a Japanese media outlet published an interview with AMD executives responsible for the company's graphics business. They were asked about the current generation of RDNA architectures, but also about what might come next.

The interview focuses on many important topics, such as primitives and mesh shaders and how they are used by PS5/Xbox consoles, but also on Multi Draw Indirect (MDI). AMD's RDNA3 implementation of MDI acceleration is already 2.3 times more efficient than the last generation. The interview does not contain any statements that would declare RDNA4 to be "X times faster" or more efficient than RDNA3. However, David Wang emphasized the importance of artificial intelligence and how it correlates with AMD's strategy going forward.

The RDNA3 architecture already provides AI acceleration through WMMA matrix multiplications (a similar block to NVIDIA Tensor and Intel's XMX Engine units). The 4Gamer editor also mentions that the RDNA3 implementation is quite modest when it comes to this unit's FP16 processing power (up to 123 TFLOPs), which is slower than the RTX 4090's 330 TFLOPs and even smaller than the Intel Arc A770's 138 TFLOPs. Needless to say, AMD still has a lot of room for improvement here.

The reason why NVIDIA is actively trying to use AI technology even for applications that can be done without using AI technology is because NVIDIA has installed a large-scale inference accelerator in the GPU. In order to make effective use of it, it seems that they are working on a theme that needs to mobilize many inference accelerators. That's their GPU strategy, which is great, but I don't think we should have the same strategy. We are focused on including the specs that users want and need to give them enjoyment in consumer GPUs. Otherwise, users are paying for features they never use. We believe that inference accelerators that should be implemented in gamers' GPUs should be used to make games more advanced and fun.

[…] The movement and behavior of enemy characters and NPCs are probably the most obvious examples [for AI cores]. Also, even if AI is used for image processing, AI should be in charge of more advanced processing. Specifically, a theme such as "neural graphics", which is currently gaining momentum in the 3D graphics industry, may be appropriate.

– David Wang, AMD

Unlike NVIDIA's inference processor, which focuses on image processing features like super-sampling technologies, AMD can achieve similar results without dedicated AI cores. However, the power of AI cores can be used in other ways. AMD sees the purpose of an AI core going beyond image processing. This kind of AI core could just as easily be used to make NPCs in games smarter. Wang claims that AI could make NPCs perform smarter tasks and more advanced tasks within the game mechanics. Although AI acceleration for game characters is probably more ambitious to implement than technologies like super-resolution or ray tracing, this could be the most interesting area where GPUs will evolve in the future.

Source: 4Gamer, Wccftech


Lade neue Kommentare

Gregor Kacknoob


524 Kommentare 442 Likes

Dabei frage ich mich schon lange, warum die vielen ungenutzten CPU Kerne nicht bereits seit Jahren zur Verbesserung der Pseudo-KI der NPCs genutzt werden. Die Hintergründe interessieren mich wirklich. Aber einen Schritt weiter gedacht frage ich mich, wie AMD darauf kommt, dass man dann dessen -Kerne für die NPCs nutzen würde.

Antwort Gefällt mir

Danke für die Spende

Du fandest, der Beitrag war interessant und möchtest uns unterstützen? Klasse!

Hier erfährst Du, wie: Hier spenden.

Hier kannst Du per PayPal spenden.

About the author

Samir Bashir