Samsung has integrated AMD’s GPU into the new smartphone processor – Techtime

by time news

January 18, 2022

Announced the Exynos 2200 chip, which includes a cluster of processors for fast tasks and a cluster of processors for heavy-duty tasks, GPU architecture brought from the world of consoles and a processor of neural networks. Started serial production in a 4 nm process

Samsung (Samsung) Announced a core chip for mobile devices that delivers graphical performance comparable to that of gaming consoles and is based on a 4-nanometer manufacturing process. The company reported that the new chip went into series production. The new Exynos 2200 chip is built in a chip-on-chip (SoC) format, and includes support for fast memories, artificial intelligence, graphical computing and fifth-generation technologies.

Its processing core is based on the concept of clusters of processors performing different tasks depending on the processing power required (similar to the new Intel processors). It features the Cortex-X2 core, which is ARM’s most powerful processor, three Cortex-A710 cores for high-performance processing, and four smaller Cortex-A510 cores for efficient tasks. For Samsung, one of the important features of the new component is the graphics processor, which provides, in its estimation, the performance of a game console.

The company has implemented the Xclipse GPU processor, which is based on AMD’s micro-architecture, RDNA 2. It includes two features that were missing until today in smartphones: Ray Tracing and Variable Rate Shading. The lighting trace provides the image with the correct illumination of each object displayed on the screen, depending on the location and intensity of the light sources in the represented environment and depending on the nature of the light reflection of each surface. Dynamic shading optimizes the work of the graphics processor by calculating the cases in which the shading hides details from the eye naturally, so there is no need to calculate them.

The visual effects are amplified by a neural network processor (NPU) at twice the power compared to the previous processor, which performs floating point calculations at a depth of 16 bits (FP16) in addition to integer calculations at a depth of 16 bits. Samsung has redesigned the Image Signal Processor architecture to support very powerful sensors that deliver images up to 200 megapixels, or HD videos that support images up to 108 megapixels that are updated 30 times per second (30fps).

Posted in categories: News, Semiconductors

Posted in tags: samsung

You may also like

Leave a Comment