**国产GPU迈向新纪元：摩尔线程推出万卡智算集群，树立AI领域新标杆**
随着人工智能的迅猛发展，算力与大模型的协同进步成为行业关注的焦点。近日，摩尔线程在上海重磅宣布其AI旗舰产品夸娥（KUAE）智算集群解决方案实现重大升级，标志着国产GPU技术迈入一个新纪元。
据了解，AI主战场中，“万卡是标配”已成为行业共识。主流大模型参数从百亿跃至1.8万亿仅耗时一年，促使各大科技巨头如META、Google、微软等纷纷搭建超大规模集群。在此背景下，摩尔线程的夸娥智算集群从千卡级别跃升至万卡规模，展现了我国在AI领域的强劲实力。
该智算集群以全功能GPU为底座，不仅具备超大规模和超强通用性，而且专为万亿参数级别的大模型训练而设计。这一创新成果为国产智算集群计算能力树立了新的里程碑，有助于推动我国人工智能领域的技术与应用创新、科研攻坚以及产业升级。
业内专家指出，摩尔线程此次升级不仅回应了算力与大模型膨胀速度的挑战，更填补了国内在超大规模、高算力智算集群的空白。此外，摩尔线程联合中国移动通信等巨头，预示着国产GPU将在AI主战场中发挥越来越重要的作用。
该产品的推出，展现了我国在人工智能领域的坚实技术实力和创新能力。随着AI技术的深入发展，期待国产GPU在未来能有更多突破与创新，推动我国AI行业的持续繁荣与进步。
英语如下：
News Title: “Local GPU Breaks Through the Limit of 10,000-Card Cluster: Moore Threads Intelligent Computing Platform Leads the New Era of AI”
Key words: Main Battleground of AI, GPU Cluster, Moore Threads
News Content: **Local GPU Steps into a New Era: Moore Threads Launches 10,000-Card Intelligent Computing Cluster and Sets a New Benchmark in AI**
With the rapid development of artificial intelligence, the coordinated progress of computing power and large models has become the focus of industry attention. Recently, Moore Threads announced a major upgrade of its AI flagship product, Kua’e Intelligent Computing Cluster solution in Shanghai, marking a new era for local GPU technology.
It is understood that in the main battlefield of AI, “10,000 cards as standard” has become an industry consensus. The mainstream large model parameters have jumped from hundreds of billions to 1.8 trillion in just one year, prompting major tech companies such as META, Google, and Microsoft to build large clusters. Against this backdrop, Moore Threads’ Kua’e Intelligent Computing Cluster has jumped from the thousand-card level to the 10,000-card scale, demonstrating strong capabilities in the AI field.
Based on fully functional GPUs, this intelligent computing cluster not only boasts ultra-large scale and strong versatility but is also designed specifically for large model training at the terabyte level. This innovative achievement has set a new milestone for the computing power of local intelligent computing clusters and will help promote technological and application innovation, scientific research breakthroughs, and industrial upgrading in China’s AI field.
Industry experts pointed out that Moore Threads’ upgrade not only responds to the challenges of the expansion speed of computing power and large models but also fills the gap in the domestic market for ultra-large-scale, high-performance intelligent computing clusters. In addition, the collaboration between Moore Threads and giants such as China Mobile shows that local GPUs will play an increasingly important role in the main battlefield of AI.
The launch of this product demonstrates China’s solid technological strength and innovation capabilities in the field of artificial intelligence. With the deep development of AI technology, we look forward to more breakthroughs and innovations in local GPUs, promoting the continuous prosperity and progress of China’s AI industry.
【来源】https://www.jiqizhixin.com/articles/2024-07-04-16
Views: 2