[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fDCWoKxanAsRcgkufw-MohE5mNBwCD26q9y3qPNgEPus":3},{"code":4,"msg":5,"data":6},200,"操作成功",{"id":7,"title":8,"content":9,"digest":10,"source":10,"coverPath":11,"thumbsCoverPath":12,"isTop":13,"isShow":14,"baseClick":13,"clickCount":15,"createTime":16,"typeId":17,"isNewest":18,"newsInfoTypeRespVo":19,"voiceUrl":22,"voiceSize":23,"taskId":24,"releaseTime":25,"titleEn":26,"contentEn":27,"voiceUrlEn":28,"taskIdEn":29,"voiceSizeEn":30},1926,"国内最大AI4S计算集群发布","\u003Cp>\u003Cstrong style=\"font-size: 18px;\">4月14日，中科曙光发布国内迄今规模最大的AI4S计算集群，并在位于郑州的国家超算互联网核心节点投入使用。该集群由6万张国产加速卡构成，标志着我国在AI4S算力+底座领域迈出关键一步。\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">北京昌平实验室里，蛋白质折叠模拟正在以前所未有的速度运行——原本需要数月甚至数年的计算，如今几天就能完成。这一突破直接加速了新药研发的关键环节，而其背后的“功臣”，不是更精密的试管，也不是更高端的显微镜，而是一个名为AI for Science（人工智能驱动的科学研究，简称“科学智能”，AI4S）的新科研范式。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\"> \u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\">AI4S，被业界视为与大语言模型、具身智能并列的人工智能三大关键方向之一，远期市场规模有望达到千亿美元级别。过去很长一段时间，中国科研人员开展AI4S研究，面临着算力紧缺、软件受限、关键工具受制于人的困境。据了解，此次发布的AI4S计算集群并不是简单的硬件堆砌，而是从底层芯片、高速互联网络到上层软件平台，全部由国内团队自主攻关，核心技术完全自主可控，能够高效处理高维函数和复杂科学问题。这意味着，科研人员可以将最核心的科研任务放心地跑在上面，不再受制于人。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\"> \u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\">有算力只是第一步。要让AI4S真正成为“换道超车”的引擎，还必须打通从算力中心到科研一线的“最后一公里”。为此，曙光同步推出了国内首个科学大模型一站式开发平台OneScience。该平台集成了数十个AI4S热点模型和数据集，用户通过国家超算互联网就能实现“开箱即用”。即使是不懂编程的生物学家、材料科学家也能快速上手，相当于为科研人员打造了一个“算力—数据—模型—实验”一体化的“超级实验室”。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\"> \u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\">目前，我国AI4S领域已在该计算集群的支撑下落地多个国家级科研场景：昌平实验室将蛋白质折叠模拟速度提升3到6个数量级，极大加快了新药研发进程；中国科学院计算所的材料筛选模型将筛选时间从数年缩短至数天；该集群还完成了414.7亿原子的超大规模模拟，首次实现了全脑860亿神经元的模拟以及万亿网格的湍流模拟，为新材料、脑科学、航空航天等领域提供了坚实的算力保障。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\"> \u003C\u002Fspan>\u003C\u002Fp>\u003Cp> \u003C\u002Fp>\u003Cp> \u003C\u002Fp>\u003Cp> \u003C\u002Fp>\u003Cp> \u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp> \u003C\u002Fp>\u003Cp>\u003Cspan style=\"color: rgb(187, 187, 187);\">【新闻来源】 北京市人民政府网 https:\u002F\u002Fwww.beijing.gov.cn\u002Ffuwu\u002Flqfw\u002Fgggs\u002F202604\u002Ft20260415_4582199.html\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"color: rgb(187, 187, 187);\">（本网转发上述文章，旨在为读者提供更多的信息资讯，所涉内容不构成投资、消费建议。文章事实如有疑问，请与有关方核实，文章观点非本网观点，仅供读者参考。）\u003C\u002Fspan>\u003C\u002Fp>","","https:\u002F\u002Fimage.51xinwei.com\u002F2026\u002F04\u002F96d00bb2854e4181b93de15cffed8f5d\u002F分组99@2x.png","https:\u002F\u002Fimage.51xinwei.com\u002F2026\u002F04\u002Fthumbs\u002F96d00bb2854e4181b93de15cffed8f5d\u002F分组99@2x.png",0,1,40,"2026-04-17 15:56",2,false,{"id":17,"name":20,"enName":21},"芯位视野","Xinwei Vision","https:\u002F\u002Fxinwei-dev-test.oss-cn-shenzhen.aliyuncs.com\u002Fintelligent\u002Faudio%3A28fc25d9-d202-41b3-af9d-98c4132bb597%3A0.wav?Expires=1776416237&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=Iz98LG2jLcUSreaujwa827gCPsA%3D",4853546,"28fc25d9-d202-41b3-af9d-98c4132bb597","2026-04-17 14:47","China’s Largest AI4S Computing Cluster Unveiled","\u003Cp>\u003Cstrong style=\"font-size: 18px;\">On April 14, Sugon unveiled China’s largest AI4S computing cluster to date, which has been deployed at the National Supercomputing Internet Core Node in Zhengzhou. Comprising 60,000 domestically developed acceleration cards, this cluster marks a critical step forward for China in the AI4S computing power + infrastructure domain.\u003C\u002Fstrong>\u003Cspan style=\"font-size: 18px;\">At the Beijing Changping Laboratory, protein folding simulations are running at unprecedented speeds—computations that previously required months or even years can now be completed in just days. This breakthrough directly accelerates a pivotal stage of new drug development. Yet its “unsung hero” is neither more precise test tubes nor more advanced microscopes, but rather a new scientific research paradigm called AI for Science (Artificial Intelligence–driven Scientific Research, abbreviated as “Scientific Intelligence,” or AI4S).\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\"> \u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\">AI4S is widely regarded by the industry as one of the three key directions in artificial intelligence—alongside large language models and embodied intelligence—with a long-term market scale projected to reach tens of billions of U.S. dollars. For an extended period, Chinese researchers conducting AI4S studies have faced severe challenges, including insufficient computing power, software restrictions, and dependence on foreign critical tools. Notably, the AI4S computing cluster unveiled this time is not merely a simple aggregation of hardware; rather, it features end-to-end domestic innovation—from underlying chips and high-speed interconnect networks to upper-layer software platforms—all independently developed by Chinese teams. Its core technologies are fully autonomous and controllable, enabling efficient processing of high-dimensional functions and complex scientific problems. This means researchers can confidently run their most critical scientific tasks on this platform without external constraints.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\"> \u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\">Having computing power is only the first step. To truly transform AI4S into an engine for “leapfrog advancement,” it is essential to bridge the “last-mile” gap between computing centers and frontline scientific research. To this end, Sugon simultaneously launched OneScience—the first all-in-one development platform for scientific large models in China. This platform integrates dozens of popular AI4S models and datasets, allowing users to achieve “out-of-the-box usability” via the National Supercomputing Internet. Even biologists and materials scientists with no programming background can quickly get started—effectively providing researchers with an integrated “super-laboratory” encompassing computing power, data, models, and experimentation.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\"> \u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\">Currently, under the support of this computing cluster, multiple national-level scientific research applications in China’s AI4S domain have already been implemented: The Changping Laboratory has accelerated protein folding simulations by three to six orders of magnitude, significantly speeding up new drug development; the Institute of Computing Technology, Chinese Academy of Sciences, reduced material screening time from years to days using its material screening model; and the cluster has successfully executed ultra-large-scale simulations involving 41.47 billion atoms—including, for the first time, full-brain simulation of 86 billion neurons and trillion-grid turbulence simulation—providing robust computing power support for fields such as new materials, brain science, and aerospace.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\"> \u003C\u002Fspan>\u003C\u002Fp>\u003Cp> \u003C\u002Fp>\u003Cp> \u003C\u002Fp>\u003Cp> \u003C\u002Fp>\u003Cp> \u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp> \u003C\u002Fp>\u003Cp>\u003Cspan style=\"color: rgb(187, 187, 187);\">[Source] Beijing Municipal People’s Government Website https:\u002F\u002Fwww.beijing.gov.cn\u002Ffuwu\u002Flqfw\u002Fgggs\u002F202604\u002Ft20260415_4582199.html\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"color: rgb(187, 187, 187);\">(This website re-publishes the above article solely to provide readers with additional information. The content herein does not constitute investment or consumption advice. Should any factual details raise questions, please verify them with the relevant parties. The viewpoints expressed in this article do not represent those of this website and are provided for readers’ reference only.)\u003C\u002Fspan>\u003C\u002Fp>","https:\u002F\u002Fxinwei-dev-test.oss-cn-shenzhen.aliyuncs.com\u002Fintelligent\u002Faudio%3Accdd8f3d-c0dc-4edd-b002-e5bbc42703d1%3A0.wav?Expires=1776652437&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=YGIcjNeAV%2BL9N9OO%2BTL9MAH58js%3D","ccdd8f3d-c0dc-4edd-b002-e5bbc42703d1",7790414]