Samsung Electronics' Exynos 2600 application processor is the first to include the AI-based ENSS feature. [Photo=Samsung Electronics website]
Samsung Electronics said it will add an AI-based graphics optimization technology called ENSS (Exynos Neural Super Sampling) to its next-generation mobile application processor, the Exynos 2600, aiming to boost performance and power efficiency.
The company said on the 28th that ENSS is software being applied to the Exynos 2600 for the first time, developed to meet increasingly demanding mobile graphics needs. It consists of NSS (Neural Super Sampling), which uses AI to reconstruct low-resolution images into higher-quality output, and NFG (Neural Frame Generation), which predicts and generates frames.
Samsung said ENSS reduces the computing load needed for screen processing, improving power efficiency while delivering clear visuals without stutter even in complex scenes. Support will be rolled out sequentially depending on product and software conditions, it said.
Samsung also said the Exynos 2600 is posting strong results on key performance indicators versus rivals. In Steel Nomad Lite, a 3D graphics benchmark measured by tech YouTubers, it showed about 15% higher performance than a competitor, the company said.
In ray tracing, a lighting technique used for more realistic visuals, Samsung said the chip recorded standout results. It ranked No. 1 on Basemark Power Board, a platform that aggregates GPU benchmark results. In one benchmark test, it scored 2,064 points, about 59% ahead of a competitor, Samsung said.
Samsung also outlined process innovation plans for the follow-up Exynos 2700. It said the Exynos 2700 will move away from a PoP (Package on Package) structure, which stacks DRAM on top of the mobile AP, and instead place the mobile AP and DRAM side by side on the same substrate. The company said the change is intended to further improve heat management, a key challenge for mobile devices.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.
