In context: Team Red confirmed off its AI-acceleration {hardware} and Radeon RX 7000 graphics playing cards in December, however did not reveal a lot about their plans — not less than not in English. Japanese media 4Gamer talked to AMD about how its AI ambitions differ from Nvidia’s.
In a just lately revealed interview, AMD executives expressed curiosity in AI functions in graphics playing cards past Nvidia’s widespread picture reconstruction. The machine translation feedback mark the primary time the corporate has mentioned the subject intimately because it unveiled its first GPU containing an AI accelerator late final yr.
Nvidia retains touting its DLSS know-how, which depends on the Tensor Cores of RTX graphics playing cards to enhance picture high quality and efficiency by machine studying. Team Red’s counterpart, FSR, strives to realize the identical purpose with out hardware-based acceleration, making it out there on extra GPUs, however could also be much less environment friendly relying on the scenario.
When AMD introduced and launched the RX 7900 XT and 7900 XTX with AI accelerators in December, onlookers thought they had been the corporate’s reply to the Team Green tensor cores. That could be the case, however AMD would not need to cease there.
David Wang, Senior Vice President of Engineering at Radeon Technologies, recommends utilizing AI acceleration to enhance the AI and motion of NPCs in video games. He additionally stated that AI might velocity up graphics pipelines by neural know-how, and talked about regular proliferation amongst different applied sciences.
AMD desires to make sure that AI acceleration in its gaming GPUs is primarily targeted on bettering the gaming expertise, Wang stated. His feedback distinction Team Red’s imaginative and prescient with Team Green’s varied non-gaming functions utilizing Tensor Cores.
Nvidia launched AI options distinctive to RTX, reminiscent of noise discount, eye contact in simulated video, and upscaling of internet video. Some of those functions do not essentially require AI, and Nvidia makes use of them to advertise its proprietary know-how, Wang stated. Wang did not criticize Nvidia’s stance, however advised that clients may pay for options they by no means use.
Wang’s interview additionally covers subjects such because the transition from primitive shaders to mesh shaders. The firm can also be engaged on know-how that might change the connection between the CPU and GPU when drawing graphics.
