Next Frontier of Supercomputer: China's Exascale in the Making

[Bussard Ramjet]

As it turns out, China is not betting solely on the Shenwei chips, and apparently has plans to build three different pre-exascale systems with three very different architectures, according to some tweets put out by James Lin, vice director for the Center of HPC at Shanghai Jiao Tong University.

3 prototype systems for exascale will be ready by end of 2017 in China. Each has ~2.5PF in Peak and ~500-600 nodes.

— James Lin@SJTU (@jameslinsjtu) July 11, 2016
The winner will be chosen to build the "exascale system" in peak performance by 2020.

— James Lin@SJTU (@jameslinsjtu) July 11, 2016
One is by NUDT with ARM approach to be deployed in Tianjing national center, where hosts Tianhe-1A.

— James Lin@SJTU (@jameslinsjtu) July 11, 2016
Another is by Taihulight team with the next generation SW CPU to be deployed on Jinan national center, where hosts Sunway bluelight.

— James Lin@SJTU (@jameslinsjtu) July 11, 2016
The third is by Sugon with AMD licenced x86 CPU to be deployed in both Shanghai supercomputer center and Shenzhen national center.

— James Lin@SJTU (@jameslinsjtu) July 11, 2016

@Bussard Ramjet

Incidentally I know the guy you are mentioning, James Lin, and I knew about this way before a thread was posted here.

 

[AndrewJin]

Incidentally I know the guy you are mentioning, James Lin, and I knew about this way before a thread was posted here.

I also know you.

 

[cirr]

China formally launches the research and development of quantum computer for the 13th 5-year period(2016-2020)

4 "deeps"

(1) deep blue(within which the quantum computer project falls)
(2) deep space
(3) deep sea
(4) deep earth

A total of 15 major programmes annouced during a press briefing by the Information Office of the State Council on 22.07.2016

国务院新闻办公室7月22日举行国务院政策例行吹风会，介绍中国创新成就与未来五年部署有关情况。

围绕深空、深海、深地、深蓝等“四深”方面的内容，科技部副部长李萌在会上介绍，在深蓝方面，我国要继续部署超级计算机，形成能力，还要启动量子计算机研发。当然这是一个很长的过程。

在深空方面，要继续实施现有的探月工程，还要部署启动首次火星探测，推进深空探测。

在深海方面，要启动深海空间站建设，加强深海探测、深海装备的关键技术研发。

在深地方面，要加强地球深部探测、城市空间安全利用、深部矿产勘探等方面。

科技部创新发展司司长许倞则介绍，深空方面，我们国家有载人航天、探月工程，载人航天“三步走”的目标是要建立空间站，在此基础上，“十三五”期间我们还在考虑启动火星探测的问题。此外在这次重大项目中，我们还在考虑空间开发利用全方面的问题，如空间飞行器在轨服务、空天地一体化网络等，我们都有一个系统的考虑。

深海方面，下一步，在深海资源勘探、国家海洋权益维护方面，我们还有大量的关键技术需要予以重点突破。其中，一个重点部署的任务就是深海空间站，要用深海空间站这样一个系统性的工程来集成提升我国深海技术和装备的系统开发与研究。除此之外，海洋重点装备研制、深海装备开发、深海传感网络开发也是我们要加强研究的重点。

深蓝，指的是网络空间、信息技术、人工智能领域的战略高技术。现在我们的超级计算机在国际上已经领先，但是现代网络空间的信息安全与维护问题，也是当前网络时代大家关心的重点和难点，也是这次15个重大项目部署的重要内容。虚拟空间的安全对于国家权益十分重要。

深地方面，不光是地球深部的矿物资源、能源资源的勘探开发，也包括在城市空间安全利用、减灾防灾等方面内容。我们对地质的了解是重要的基础性工作，浅层一点关乎到城市开发中地下管网、基础设施建设，深层关乎国家重要战略资源和能源开发利用，所以这项工作对于国家资源开发、城镇化建设都非常重要。

@Bussard Ramjet

 

[ChineseTiger1986]

TIANJIN, July 26 (Xinhua) -- Scientists have started to develop a sample machine that will play a part in the development of a new era supercomputer capable of a billion-billion calculations per second, researchers said Tuesday.

The National Supercomputer Center in Tianjin is developing the exascale supercomputer with the National University of Defense Technology (NUDT), said Meng Xiangfei, assistant director of the center.

The aim is to make the computer, which will be capable of a quintillion calculations per second, by 2020.

It will be 200 times faster than China's Tianhe-1 supercomputer, and even faster than the most powerful supercomputers in the world, he said.

China's Sunway TaihuLight is the world's fastest system, capable of 93 quadrillion calculations per second, according to an announcement at the 2016 International Supercomputing Conference, Germany.

This new project will expend on existing theories and applications of the exascale system's hardware and software. The sample machine will hopefully be ready by early 2018, Meng said.

The new era computing system will mark a huge advancement in terms of intensity of calculation, capacity of single chips and the rate of data transmission, he said.

In 2010, the country's first petaflop supercomputer, Tianhe-1, which is capable of at least a million billion calculations per second, was unveiled to the world.

At present, Tianhe-1 supports various tasks including oil exploration, high-end equipment manufacturing, biological medicine and animation design, and serves nearly 1,000 customers.

http://news.xinhuanet.com/english/2016-07/26/c_135542281.htm

 

[war&peace]

China's development model is better than USSR on electronics and supercomputing side and that's why USSR/Russia has been behind the west in electronics and modern avionics.

 

[ChineseTiger1986]

China's development model is better than USSR on electronics and supercomputing side and that's why USSR/Russia has been behind the west in electronics and modern avionics.

Russia is also learning the mistakes from the past, that's why they are trying to catch up.

https://www.hpcwire.com/2015/09/28/t-platforms-leads-russias-top-50-supercomputing-list/

 

[Beny Karachun]

I cant understand why people develop super computers today, rather than developing the chips to be more power efficient and powerful and later make them?
They cost millions to make and millions to keep and millions to operate.

 

[cirr]

I cant understand why people develop super computers today, rather than developing the chips to be more power efficient and powerful and later make them?
They cost millions to make and millions to keep and millions to operate.

People develope power efficient chips and powerful chips before developing supercomputers, ie chips used in supercomputers are super power efficient and powerful.

 

[Providence]

China is doing some great innovations in supercomputer space. I hope their algorithms are powerful enough to make use of it ..

 

[艹艹艹]

China starts developing new-era exascale supercomputer
(Xinhua) 01:13, July 27, 2016

Scientists have started to develop a sample machine that will play a part in thedevelopment of a new era supercomputer capable of a billion-billion calculations persecond, researchers said Tuesday.

The National Supercomputer Center in Tianjin is developing the exascale supercomputerwith the National University of Defense Technology (NUDT), said Meng Xiangfei, assistantdirector of the center.

The aim is to make the computer, which will be capable of a quintillion calculations persecond, by 2020.

It will be 200 times faster than China's Tianhe-1 supercomputer, and even faster than themost powerful supercomputers in the world, he said.

China's Sunway TaihuLight is the world's fastest system, capable of 93 quadrillioncalculations per second, according to an announcement at the 2016 InternationalSupercomputing Conference, Germany.

This new project will expend on existing theories and applications of the exascale system'shardware and software. The sample machine will hopefully be ready by early 2018, Mengsaid.

The new era computing system will mark a huge advancement in terms of intensity ofcalculation, capacity of single chips and the rate of data transmission, he said.

 

[Beny Karachun]

People develope power efficient chips and powerful chips before developing supercomputers, ie chips used in supercomputers are super power efficient and powerful.

What are these super computers capable of?
Or what's the reason to build them?

 

[xunzi]

It is good see how far we can push the technological edge of supercomputer. This is the moment our Computer Scientists (one of my old professor is a Chinese computer scientist) to demonstrate what they can do and we should offer full support to their imagination.

 

[cirr]

What are these super computers capable of?
Or what's the reason to build them?

I for one am interested in learning the early results of applying Taihulight to the simulation and design of next generation of nuclear weapons.

 

[Beny Karachun]

I for one am interested in learning the early results of applying Taihulight to the simulation and design of next generation of nuclear weapons.

I am pretty sure there is no reason to design any because China, the US and Russia already have enough to destroy the world a few times.

 

[cirr]

I am pretty sure there is no reason to design any because China, the US and Russia already have enough to destroy the world a few times.

Yes for the US and Russia but no for China.

China needs to stock up on its nuclear arsenals at least 10 times to stop any country from having any silly intent on China.

Mind you, miniturized nuclear warheads are perfect for a variety of newly developed delivery vehicles such the DF-ZF and DF-26.