


In less than half a year, the world’s most authoritative supercomputer TOP500 list has been updated again!
On November 14, at the SC22 conference held in Dallas, USA, the new issue of "The World's Fastest 500 Supercomputers" was announced.
Compared with the June issue, the top 10 in the new issue only have slight adjustments, and the top 3 remain unchanged.
Frontier from Oak Ridge National Laboratory (ORNL) still tops the list with 1.102 EFlop/s, and its HPL score is almost three times that of the second place. This is a major victory for computer science, and it is also the only "Exascale supercomputer" on this issue's TOP500 list.
"Fugaku" of the Riken Center for Computational Science (R-CCS) in Kobe, Japan, had two consecutive times before being squeezed out by the Frontier machine. year to occupy the top spot. This time the list is still ranked 2nd.
LUMI, "Europe's fastest supercomputer", which ranked third in the previous list, also successfully retained the top three this time s position. However, LUMI has undergone a major upgrade this time, doubling the size of the machine and achieving an HPL score of 0.309 EFlop/s.
The only new machine in the top ten this time is the Leonardo system of EuroHPC/CINECA located in Bologna, Italy. The machine achieved an HPL score of 0.174 EFlop/s with 1,463,616 cores, ranking 4th.
Due to the strong rise of Leonardo, compared with the 6th and 9th places in the previous period, China’s two supercomputers "Sunway·TaihuLight" ” and “Tianhe 2A” were moved down one place to 7th and 10th.
Since 2017, the speed of improvement in supercomputing power has begun to slow down, resulting in the "E-class supercomputing" that has been talked about for several years until last year. This may indicate that Moore's Law is ending.
AMD is still yes, NVIDIA GPU "seeks to fail alone"
In the field of supercomputer CPUs, as the first choice for HPC systems, AMD's advantages have become increasingly obvious.
According to statistics, although only 20.2% of the machines in this Top500 use AMD X86 processors and 75.8% of the machines are based on Intel X86 processors, half of the new systems and 75 % of new performance - all relying on AMD chips.
Among them, Frontier uses Gen AMD EPYC processors optimized for HPC and AI, as does the third-ranked LUMI system.
It is expected that as time goes by, the number of supercomputers using AMD chips will continue to rise, while Intel’s share will continue to decline.
In terms of GPU, the competition is between Nvidia and AMD.
With the emergence of Frontier "copycats" in Europe and the United States, AMD has also gained some quite important shares in the high-end market.
Currently, the most watched supercomputer in the world is Argonne National Laboratory’s all-Intel Aurora supercomputer—a supercomputer with twice the performance of Frontier.
However, the machine has been delayed since 2018 due to Intel's failure to deliver the chips on time. In September, Intel announced that it had begun delivering its Sapphire Rapids CPU and Ponte Vecchio GPU blades to Argonne for integration into the system.
In the newly updated TOP500 list, we did not find this "Thanos" supercomputer.
China: ranked first in quantity and only fourth in computing power
Although it is at a disadvantage in terms of single unit performance, after this issue of the TOP500 list, China It is still called "the biggest unknown" by foreign media.
In this issue of the list, although China only has two supercomputers, "Sunway TaihuLight" and "Tianhe-2", among the top 10, there are a total of 500 supercomputers on the list. , China occupies 162 units, which is down from 173 units last year, but still 31 units more than Europe and 36 units more than the United States, firmly ranking first in the world.
In terms of computing power, the United States ranks first with 43.2% of deployed computing power, Japan ranks second with 19.6%, and China ranks fourth with 10.6%.
It is worth noting that China has stopped submitting the most advanced supercomputing system information to the TOP500 committee, which has also led to a decline in the global supercomputer When counting the numbers - foreign media had to exclude China's data.
Last year, foreign media reported that China is likely to have developed an E-class supercomputer earlier than the United States, and may even have more than one.
The first one is the 1.3 exaflops "Sunway Ocean Light" of the Wuxi National Supercomputing Center. The system was tested using the High-Performance Linpack (HPL) benchmark in March 2021 and achieved a sustained performance of 1.05 exaflops;
The second one is located at the Guangzhou National Supercomputer Center Tianhe-3 system. It is also said to have reached a peak capacity of 1.3 exaflops, with sustained performance of over 1 exaflops in HPL testing.
However, neither of these two machines appears in the Top500 rankings from 2021 to now.
Earlier this year, the United States banned the export of AMD Instinct and Nvidia's A100 and H100 GPUs to China and Russia for exceeding newly mandated performance limits. Nvidia is even going so far as to weaken its A100 in order to continue supplying China with chips.
The above is the detailed content of The world's top 500 supercomputers are released! One year after 'abandoning the exam”, China's number of supercomputers is still far ahead. For more information, please follow other related articles on the PHP Chinese website!

本站1月30日消息,意大利能源巨头Eni埃尼集团近日宣布将建设全新的HPC6超级计算机。该超算将基于AMDEPYCCPU和InstinctMI250XGPU打造,完成后理论峰值算力将达600PFlop/s,持续峰值算力也将有400PFlop/s,可在现有Top500榜单中排到第五。经查询得知,埃尼集团是世界传统大型石油公司之一,同时也是意大利最大的三家企业之一。据了解,该公司由意大利政府持股约30%。▲上一代HPC5超算照片。图源埃尼集团官网埃尼集团现有2座超算进入了Top500超算榜单:之前的

本站6月26日消息,全球最强超级计算机所配备的超强GPU数量不到5万片,而AMD公司为了增强和英伟达的竞争力,目标推出支持120万片GPU的超级数据中心AI集群。在接受TheNextPlatform采访时,AMD数据中心解决方案集团执行副总裁兼总经理Forrest Norrod表示,公司正在认真考虑构建最大规模的人工智能GPU集群。本站注:人工智能训练集群通常由数千片GPU构建,通过高速互连连接服务器机架,而AMD正研究的120万片无疑在数量级上指数倍增长,这必然带来延迟、功耗、硬件故障等多方面

本站4月25日消息,英伟达公司近日宣布和日本产业技术综合研究所(AIST)合作,搭建名为“ABCI-Q”的超级计算机,将整合传统超级计算机和量子计算机打造出混合云系统。由于英伟达表达单独运行量子计算器仍然会犯很多错误,因此超级计算机必须帮助解决错误,让复杂的运算变得更加顺畅。ABCI-Q+%执行高速、复杂计算的能力将有助于人工智能、能源和生物技术领域的研究和企业应用,例如提高新药开发和物流的效率率。该网站从报道中获悉,ABCI-Q+内置超过2000片英伟达H100TensorCoreGPU,并通

戈登贝尔奖(ACMGordonBellPrize)设立于1987年,由美国计算机学会颁发,被称为超算界的「诺贝尔奖」。该奖项每年颁发一次,用以表彰高性能计算领域取得的杰出成就。奖金1万美元,由高性能和并行计算领域先驱戈登·贝尔提供。近日,在全球超级计算大会SC23上,2023年ACM戈登贝尔奖授予了美国和印度研究人员组成的8人国际团队,他们实现了大规模量子精度的材料模拟。相关项目名称为「量子精度的大规模材料建模:金属合金中准晶体和相互作用扩展缺陷的从头计算模拟」。团队成员的背景各不相同,他们分别

半年不到,全球最权威的超级计算机TOP500榜单又更新了!11月14日, 在美国达拉斯举行的SC22会议上,新一期「世界上最快的500台超级计算机」重磅公布。和6月那期相比,新一期榜单前10名仅有微调,前3名都没有变化。美国橡树岭国家实验室(ORNL)的Frontier仍以1.102 EFlop/s位列榜首,其HPL分数几乎是第二名的三倍。这对计算机科学来说是一个重大胜利,也是本期TOP500榜单中唯一的「E级超算」。日本神户Riken计算科学中心(R-CCS)的「富岳」(Fugaku)在被Fr

已经霸榜两年,完成四连冠的日本富岳终于跌落了「神坛」。 在刚刚公布的2022年全球超算Top500名单中,美国橡树岭国家实验室(ORNL)的Frontier夺得第一。 凭借1.102 Exaflop/s的HPL分数,Frontier不仅是有史以来最强大的超级计算机,它也是第一台真正的「E级超算」。 相比之下,中国的神威·太湖之光和天河二号排名下滑至第六和第九。 去年,美国将神威·太湖之光的处理器供应商申威列入了实体清单,或多或少给中国超算的发展造成了影响。 这次Top500的64位浮点计算

重写后的内容:来源图片:摄图网新华社消息,英国政府11月2日宣布,将对人工智能(AI)研究资源的投资增至3亿英镑(约合3.66亿美元),这比今年3月宣布的1亿英镑加码了两倍,以使英国超级计算能力提高30倍以上。英国科学、创新和技术大臣米歇尔·唐兰2日在英国布莱奇利园举办的人工智能安全峰会上宣布,将在英国建造并连接两台新的超级计算机,使研究人员未来访问资源时所借助的英国公共人工智能计算设施计算能力提升30倍以上。这两台超级计算机连接后将于2024年夏季开始运行。研究人员可使用这些机器分析先进的人工

英伟达A100被禁的消息才刚刚过去没多久,美国新的制裁又来了……据知情人士称,拜登政府计划在下个月扩大针对中国芯片制造领域的制裁。一句话概括就是,禁止向生产14nm以下先进半导体的中国工厂出口芯片制造设备,除非获得商务部的许可。一剑封喉据说,美国商务部正在酝酿,依据之前给三家美国公司下发的信函中提到的限制措施,发布一套全新的规定。消息人士称,这些规定可能会包括针对中国的额外行动。其中,这三家美国公司为KLA(科磊)、LamResearch(拉姆研究)和AppliedMaterials(应用材料公


Hot AI Tools

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Undress AI Tool
Undress images for free

Clothoff.io
AI clothes remover

AI Hentai Generator
Generate AI Hentai for free.

Hot Article

Hot Tools

SecLists
SecLists is the ultimate security tester's companion. It is a collection of various types of lists that are frequently used during security assessments, all in one place. SecLists helps make security testing more efficient and productive by conveniently providing all the lists a security tester might need. List types include usernames, passwords, URLs, fuzzing payloads, sensitive data patterns, web shells, and more. The tester can simply pull this repository onto a new test machine and he will have access to every type of list he needs.

MinGW - Minimalist GNU for Windows
This project is in the process of being migrated to osdn.net/projects/mingw, you can continue to follow us there. MinGW: A native Windows port of the GNU Compiler Collection (GCC), freely distributable import libraries and header files for building native Windows applications; includes extensions to the MSVC runtime to support C99 functionality. All MinGW software can run on 64-bit Windows platforms.

SublimeText3 Mac version
God-level code editing software (SublimeText3)

SAP NetWeaver Server Adapter for Eclipse
Integrate Eclipse with SAP NetWeaver application server.

Zend Studio 13.0.1
Powerful PHP integrated development environment
