在今日的英伟达GTC 2024大会上，全球科技巨头英伟达的CEO黄仁勋揭晓了备受瞩目的新一代GPU——Blackwell。首枚Blackwell芯片GB200的发布，标志着英伟达在图形处理技术上的又一重大突破，预计将于今年晚些时候正式上市，为业界带来更强大的计算性能。
值得注意的是，SK海力士作为英伟达的唯一HBM3内存供应商，紧随其后宣布了一个重要消息。该公司透露，已经开始量产其最先进的高带宽内存产品HBM3E，并计划从本月下旬起向全球客户供货。这一举措无疑为Blackwell GPU的性能提升提供了坚实的硬件支持。七个月前，SK海力士曾公开宣布成功开发出HBM3E，此次量产标志着该技术从研发阶段正式迈入商业化应用。
HBM3E内存的量产和供货，不仅将提升英伟达Blackwell GPU的数据处理速度，也将对数据中心、人工智能、高性能计算等领域产生深远影响。此次合作凸显了两家公司在技术创新和供应链协作上的紧密关系，预示着未来高性能计算领域可能出现更多里程碑式的进展。
英语如下：
News Title: “SK Hynix Begins Mass Production of HBM3E Memory, NVIDIA’s Next-Gen GPU Blackwell Set to Feature It”
Keywords: SK Hynix, HBM3E, NVIDIA GPU
News Content: At the NVIDIA GTC 2024 conference today, tech giant NVIDIA’s CEO, Jensen Huang, unveiled the much-anticipated next-generation GPU, the Blackwell. The launch of the GB200 chip marks another significant milestone in NVIDIA’s graphics processing technology, with an official release scheduled later this year, promising to deliver enhanced computational power to the industry.
Notably, SK Hynix, NVIDIA’s sole supplier of HBM3 memory, followed up with an important announcement. The company disclosed that it has commenced mass production of its most advanced High Bandwidth Memory (HBM3E) product and plans to supply global customers starting from the latter part of this month. This development solidifies the hardware foundation for the performance boost of the Blackwell GPU. Seven months ago, SK Hynix publicly announced the successful development of HBM3E, and this mass production signifies the technology’s transition from the research and development phase to commercial application.
The mass production and supply of HBM3E memory will not only enhance the data processing speed of NVIDIA’s Blackwell GPU but also have a profound impact on sectors such as data centers, artificial intelligence, and high-performance computing. This collaboration underscores the close relationship between the two companies in terms of technological innovation and supply chain collaboration, foreshadowing potential future milestones in high-performance computing.
【来源】https://www.cls.cn/detail/1623431
Views: 1