据最新报道，全球科技巨头三星电子正加速推进其人工智能（AI）芯片创新，计划在年底前向韩国知名IT企业Naver交付自主研发的AI芯片Mach-1。这笔交易预估价值最高可达1万亿韩元，约合54.2亿元人民币，彰显了三星在AI芯片领域的雄心壮志。
三星电子DS部门的负责人庆桂显透露，Mach-1芯片目前正处于SoC设计的最后阶段，公司目标是今年底完成生产。这款非传统结构的AI芯片设计独特，能够将核心计算芯片与LPDDR片外内存之间的数据传输瓶颈降低至现有AI芯片的八分之一，同时在能耗上仅为英伟达同类产品的八分之一，实现了性能与能效的双重突破。
据悉，三星与Naver自2022年末开始展开深度合作，共同研发AI芯片，旨在为Naver的超大规模AI模型HyperCLOVA提供定制化解决方案。此次合作的成果Mach-1芯片，有望大幅提升Naver在AI领域的计算效率，进一步推动其在人工智能应用上的领先地位。
这一交易不仅体现了三星在半导体技术上的领先地位，也预示着未来AI芯片市场将可能出现更多高性能、低能耗的产品，对整个行业产生深远影响。
英语如下：
News Title: “Samsung Electronics’ Year-End Gift to Naver: The AI Chip Mach-1, High Performance, Low Power, Kicking Off a New Era of Intelligence!”
Keywords: Samsung AI Chip, Naver Collaboration, Mach-1 Transaction
News Content:
Samsung Electronics is preparing to deliver its self-developed AI chip, Mach-1, to South Korean IT giant Naver by the end of the year in a deal estimated to be worth up to 1 trillion Korean won, approximately 54.2 billion Chinese yuan. This move highlights the tech behemoth’s ambitious plans in the AI chip sector.
According to recent reports, Samsung’s Device Solutions (DS) division head, Kyung Kye-hyun, revealed that the Mach-1 chip is in the final stages of System-on-Chip (SoC) design, with a target to complete production by the end of 2023. The unconventional AI chip is uniquely designed to reduce data transfer bottlenecks between the core processing chip and LPDDR external memory to one-eighth of current AI chips, while consuming only one-eighth the power of comparable NVIDIA products, achieving a dual breakthrough in performance and efficiency.
Sources indicate that Samsung and Naver began collaborating deeply on AI chip development since the end of 2022, aiming to provide a customized solution for Naver’s large-scale AI model, HyperCLOVA. The outcome of this partnership, the Mach-1 chip, is anticipated to significantly enhance Naver’s AI computing efficiency, further solidifying its leading position in AI applications.
This transaction not only underscores Samsung’s semiconductor prowess but also foreshadows the emergence of more high-performance, low-power AI chips in the future, with potential far-reaching impacts on the entire industry.
【来源】https://www.ithome.com/0/757/537.htm
Views: 1