“RDNA 4?” You cry, “but RDNA 3 just came out!” Indeed, and the release of AMD’s Radeon RX 7900 XT and RX 7900 XTX graphics cards just two months ago marked the debut of AMD’s third iteration of its RDNA graphics architecture. Nevertheless, AMD is already talking about RDNA 4. In an interview with the Japanese site 4Gamer, AMD gave impressions of a few critical details about its next-generation GPUs.
The comment didn’t come out directly, saying that “RDNA will be 4 [x]” or have [x]”, but rather as statements about AMD’s David Wang and Rick Bergman’s ideas about how computer graphics will evolve. Wang, for example, acknowledged that NVIDIA has put a lot of emphasis on using AI. Wang says AMD doesn’t have that strategy , and that AMD doesn’t think GPU AI accelerators will be well used in the consumer market.
It’s a bit of an odd statement considering that RDNA 3 actually added AI inference acceleration to the architecture for the first time, but it makes sense given that even an Arc Alchemist GPU will outperform a Radeon RX 7900 XTX in pure tensor operations can . With that in mind, it might follow that AMD would be dismissive of AI in the consumer market, but that’s not exactly what Wang was saying. Can you tell what FSR or DLSS is? (Clockwise from left: Native, FSR1, FSR2, DLSS)
Instead, he said AMD is focused on “incorporating the specs that users want” to deliver the best gaming experience. Wang notes that users otherwise pay for features they don’t use. He points to the example of NVIDIA’s DLSS, where newer versions of AMD’s FidelityFX Super Resolution offer a relatively similar experience without using dedicated hardware.
Wang says that instead of image processing, he would prefer to see AI acceleration on graphics cards used instead to make games “more advanced and fun,” citing the example of “the movement and behavior of enemy characters and NPCs.” . an area of game programming often referred to as “AI” despite its distant relationship to the study of neural networks.
In response to a question about “new 3D graphics paradigms” that need to be standardized in the graphics industry, AMD noted that the next big step for graphics actually has to do with “GPU independent drawing” that takes the CPU overhead from Graphics tasks eliminated. We’ve already taken a step in this direction with newer DirectX 12 and Vulkan graphics APIs, but it sounds like AMD is interested in pushing this idea further.
AMD’s only mention of the MDIA was a tiny note on this slide.
4Gamer points out that AMD’s RDNA 3 includes a “Multi-Draw Indirect Accelerator” in hardware, which is used for indirect multi-draw operations. This is a method of distributing multiple instantiated drawing commands so that they can be issued entirely on the GPU, saving quite a bit of overhead. Apparently, RDNA 3 can do this 2.3 times faster than RDNA 2, and we might see further improvements in this direction in RDNA 4.
Probably the most interesting part of the interview, at least in relation to RDNA 4, is the statement towards the end where AMD’s Rick Bergman says that the company “promises to evolve towards even higher performing RDNA 4 in the near future”. It shouldn’t surprise anyone that the company expects its next-gen GPUs to be even faster than its existing products, but “in the near future” sounds like the company could move to RDNA 4 sooner than expected.