During IFA 2024, Jack Huynh, Senior Vice President and General Manager of AMD's Compute and Graphics Business Unit, was interviewed by TomsHardware and shared insights about AMD's forthcoming plans in the graphics domain. Notably, he confirmed a strategic shift for Radeon graphics cards to focus more on market size rather than flagship gaming GPUs.

According to TomsHardware, Jack Huynh also divulged other significant details about future GPUs and related graphics technologies. One of the key highlights was the discussion on AMD FidelityFX Super Resolution 4 (FSR 4).

Jack Huynh touched upon the growing demand for handheld gaming devices, emphasizing that AMD's top priority is extending battery life. For example, in ASUS ROG handhelds, AMD aims to enable gamers to play titles like Black Myth: Wukong for 3 to 4 hours continuously, compared to the current 60-minute battery life. This is where Frame Generation and Interpolation technology come into play, which will be one of the core features of FSR 4. Unlike FSR 2 and 3, which were based on analytics and filtering techniques, FSR 4 will fully transition to AI-based technology.

The main reason for the previous approach in FSR 2 and 3 was to bring the product to market swiftly. However, about 9 to 12 months ago, AMD's team pivoted to an AI-driven approach to maximize efficiency, extending battery life and securing a stable frame rate of around 30fps for handheld devices. One of the remarkable features of past FSR versions has been their compatibility with a broad array of GPUs, and an AI-based solution will not inherently limit the use of GPUs for computational tasks.

Intel's XeSS, an AI-based super-resolution technology, leverages machine learning to upscale low-resolution frames to high-resolution using XMX AI acceleration in the Xe core of the Xe-HPG architecture, thus boosting frame rates. XeSS is also fully compatible with DirectX 12, extending AI-based super-resolution capabilities to various GPUs, including integrated graphics cards, through the DP4a instruction set. Hence, it is plausible that FSR 4 will adopt a similar AI-based approach, enhancing quality while optimizing energy efficiency and ensuring broad compatibility.