It is reported that Samsung independently developed ray tracing and AI oversampling technology, which will be used in Exynos chips after 2025.
CTOnews.com, November 7 (Xinhua) although Samsung has been working with AMD for the past few years to bring ray tracing to its Exynos chip, it has recently been reported that Samsung seems to be developing its own ray tracing and AI supersampling technology, which is planned to be used on future Exynos chips. CTOnews.com noted that just a few days ago, it was reported that Samsung was working with AMD and Qualcomm to introduce FSR (FidelityFX Super Resolution) into its phones.
According to Daily Korea, a team at the Samsung Institute of Advanced Technology (SAIT) appears to be working on two new technologies: neural light reconstruction and neural supersampling. These technologies are expected to be used in Exynos chips after 2025. If this report is accurate, these gaming technologies may not be included in Samsung's 3nm "Dream Chip" for the Galaxy S25 series. The report says Samsung's ray-tracing technology is better than AMD and Nvidia's because it can ray-trace all parts of the scene. By contrast, the ray tracing technology of AMD and Nvidia is usually applied only to the part of the game screen that reflects light.
Because ray tracing requires a lot of processing power, Samsung uses NPU (neural processing unit) for ray tracing related calculations and uses an AI-based oversampling method (neural oversampling). This method is similar to AMD's FSR and Nvidia's DLSS, which can render the picture at a lower resolution and then zoom in to a higher resolution to reduce the burden on the chip. After that, the edges of the objects in the picture will be anti-aliased to make them smoother. Since these technologies have not yet been officially released, it is important to be cautious about this information.
Samsung is expected to launch the Galaxy S24 in mid-January 2024, which will be the company's first AI phone with a powerful NPU chip (Exynos 2400 or Snapdragon 8 Gen 3) that can provide AI processing on the device side without the Internet. In addition, the company plans to enhance the performance of NPU to provide ray tracing and supersampling tasks for future Galaxy smartphones.
