Samsung Electronics, which aims to be the No. 1 non-memory device manufacturer until 2030, is spurring the development of its own graphics processing unit (GPU). GPU is a device specialized for processing high-volume data like graphics. It is an essential device in the era of artificial intelligence (AI) and autonomous driving.
According to recent employment announcements by the Samsung Austin R&D Center (SARC), Samsung Electronics is recruiting employees with expertise in GPU, aiming to develop a product that can be installed in its flagship smartphone Galaxy S series, contrary to the expectations that it would first install its self-developed GPU on low-end smartphones to verify its performance.
There was a rumor last year that Samsung would launch a gaming smartphone equipped with its own GPU, but it is clear now that the GPU will be first used for the company’s flagship Galaxy S series. Samsung's GPU development is led by Dr. Chien-Ping Lu from NVIDIA, who joined last year.
In today's mobile environment where large amounts of data like 3D graphics must be processed quickly, GPU is an integral part needed to enhance the performance of application processors (APs). ARM's GPU Mali is currently installed on Samsung’s Exynos AP, but it does not stack up against the GPUs from Apple and Qualcomm. Some critics also pointed out that ARM’s GPU takes up too much space in an AP, making it difficult to improve power efficiency and performance. Samsung’s contract with ARM is known to expire next year.
Samsung’s employment announcement stated that “the next generation GPUs will cover growing markets such as autonomous driving, machine learning, and deep learning." GPU is indispensable for automotive semiconductors, which Samsung Electronics focuses on as a next-generation business. GPU processes a large number of images that a vehicle acquires through various sensors and figures out how to move the vehicle through machine learning and deep learning.
The automotive semiconductor market, which is dominated by NVIDIA, is expected to grow rapidly with the development of autonomous navigation technology. Market research firm IHS Markit predicted that the automotive semiconductor market will grow 35.8 percent from US$25.6 billion last year to $US35.5 billion in 2023.
GPU is a type of processor that processes graphics images on a computer to allow them to be viewed on the screen. As the graphics quality of multimedia content becomes more sophisticated and complex, GPU has evolved to maximize the computation performance to process image data faster. In particular, the recent GPU technology has attracted attention for its application in the AI field. Alpha Go, which won the five-game Go match against human in March of 2016, was also developed based on the AI learning ability based on accelerated GPU computation. Major GPU manufacturers include NVIDIA, AMD and Intel.