New AMD update releases FSR 4.1 system for Radeon RX 9000 cards and improves AI on PS5 Pro

Semiconductor manufacturer AMD recently released the Radeon Software Adrenalin Edition 26.3.1 driver, officially introducing support for FSR 4.1 upscaling technology. Esta software update hits the market exclusively for Radeon RX 9000 series graphics cards, marking an important transition in the way the company handles graphics rendering. The new driver package strategically coincides with the arrival of major titles on the electronic games market, using these works as a technological showcase. The main innovation of this version lies in the intensive application of machine learning algorithms for image reconstruction. Diferente From previous iterations, which relied on traditional spatial and temporal methods, the new system uses neural networks to fill in missing visual information. Esse movement aligns the company with current trends in the hardware industry, where artificial intelligence takes the central role in optimizing performance. The launch takes place in parallel with the improvements implemented in the PSSR system of the Sony console. The engineering behind these new features reflects months of research and development focused on delivering greater visual fidelity without compromising the frame rate per second. Users of desktop computers equipped with compatible hardware can now download the update through the manufacturer’s official channels.
The update package brings immediate support for specific games that integrate the new technology from the source code. Entre the main compatibility additions confirmed by the hardware developer include: – Suporte native to the recently released Crimson Desert. – Otimização dedicated to Death Stranding 2: On the Beach. – Melhorias in overall operating system stability when running graphics intensive applications.
The exclusivity for the brand’s latest architecture generates discussions among technology consumers. The company justifies the decision based on the matrix processing requirements that only new generation artificial intelligence accelerators have. Especialistas in hardware point out that this segmentation represents a generational leap necessary for the evolution of the graphics ecosystem.
Technical details of the new upscaling architecture
The updated version refines image processing by acting directly on three main rendering fronts. The first notable aspect involves the richness of detail in elements of high geometric complexity, such as dense foliage, character hair and irregular surface textures. The neural network trained by the manufacturer can interpret these patterns with greater precision, avoiding the blurring effect common in older upscaling techniques. The treatment of visual artifacts received special attention from the software engineering team. The practical result delivers a final image that, in many scenarios, approaches native resolution, requiring considerably less computational power from the graphics processing unit.
Another point of technical evolution focuses on temporal stability during sudden camera movements. Fast-action Jogos often suffer from visual trailing or flickering when the algorithm cannot accurately predict upcoming frames. The new artificial intelligence model analyzes motion vectors with greater efficiency, ensuring smooth transitions even in highly dynamic scenes. The configuration mode focused on maximum performance manages to maintain acceptable image quality, allowing high refresh rate monitors to be fully utilized without sacrificing visual clarity.
Integration of Machine Learning in Recent Games
The title Crimson Desert serves as the main demonstration environment for the capabilities of the new graphics technology. The digital entertainment work incorporates simultaneous support for multiple optimization tools, including previous iterations of the upscaling system and light ray reconstruction technologies. Native implementation allows the game’s graphics engine to communicate directly with the graphics card’s artificial intelligence cores.
The denoising functionality for ray tracing presents expressive results in rendering reflections and global illumination. The algorithm cleans up visual noise generated by real-time ray tracing, delivering sharper reflective surfaces and physically accurate shadows. The combination of these technologies creates a highly detailed virtual environment with a manageable computational cost.
The availability of these tools from the first day of the game’s launch demonstrates close alignment between the hardware manufacturer and development studios. Early integration makes code optimization easier and ensures consumers have access to the best possible visual experience immediately after installing the software.
Joint development at Projeto Amethyst
The algorithmic basis of the new computer upscaling system shares its neural network with the technology present in the Sony desktop console. Essa technical collaboration, known behind the scenes in the industry as Projeto Amethyst, began in research laboratories in 2023. The main objective of the partnership involves the creation of highly efficient artificial intelligence models for graphic enhancement on different platforms.
Working together allows both companies to save research resources and accelerate the implementation of innovations in the consumer market. The version of the technology that reaches the console goes through an additional period of refinement, adapting perfectly to the fixed hardware specifications. Esse extra development time results in highly optimized code for the unified home entertainment system architecture.
The console operating system update introduced the improved version of the image reconstruction tool for users. Recent Títulos, such as Resident Evil Requiem, already use the new version of the algorithm to deliver sharper graphics in high performance modes. A switch in system settings allows players to activate the improvement in games that already supported the original version of the technology.
Representatives from the console manufacturer confirm that the technology developed in partnership delivers measurable advances in image stability and visual clarity. Standardizing certain artificial intelligence instructions between desktop computers and consoles makes it easier for game developers to apply similar optimizations across multiple platforms with less programming effort.
Data processing and hardware differences
Despite sharing the same algorithmic origin and fundamental neural network concepts, the technology’s implementations are rigorously adapted to the physical realities of each hardware. The Radeon RX 9000 series graphics cards utilize dedicated processing units exclusively to accelerate complex matrix instructions, operating independently of traditional rendering cores. Essa physical separation on the chip allows upscaling to occur simultaneously with graphics generation, reducing overall system latency. In the desktop console environment, the architecture requires a different approach due to the design of the main processor. The Sony equipment performs artificial intelligence calculations using conventional computing units, but with specific acceleration in INT8 data format. Essa configuration achieves a computational performance of up to 300 operations per second, which is sufficient to process image reconstruction in real time. Análises independent techniques point out that the adaptation for the console considers architectural limitations inherent to the format, such as the absence of a dedicated level 3 cache and the use of a unified memory pool. On desktop computers, the presence of separate video memory and high-speed caches allows data to be trafficked more aggressively, justifying the absolute performance differences between platforms.
Restrictions on use in RDNA 4 generation
The official documentation that accompanies the new video driver keeps the upscaling technology restricted to the manufacturer’s most recent architecture. Não there are mentions about the expansion of official support for the RX 6000 or RX 7000 generation product lines. The lack of backward compatibility raises technical questions in specialized forums and hardware discussion platforms.
Independent technology professionals argue that previous generation video cards have the theoretical capacity to execute the necessary instructions in INT8 format. The semiconductor manufacturer chooses not to publicly comment on the exact reasons for the software block, maintaining its exclusive focus on promoting the new line of graphics accelerators.
Mod community movement
The official restriction imposed by the software did not stop technology enthusiasts from looking for alternative solutions. Grupos of software modifiers are already reporting success in unofficially running the new technology on video cards based on the RDNA 2 and RDNA 3 architectures.
Graphic evolution in the computer and console market
Continuous technical development among hardware manufacturers indicates a standardization in the use of neural networks for rendering electronic games. Collaboration directly benefits the development ecosystem, creating robust tools that raise the average visual quality of releases. Jogadores on both platforms experience qualitative leaps in image sharpness without the need to invest in extremely heavy native resolutions.
The technological advancement documented in this update significantly reduces the gap with proprietary solutions offered by competing companies in the graphics card industry. The digital entertainment market is moving towards a scenario where artificial intelligence dictates the performance rules, making raw processing power just one part of the graphic equation.






