Scientists have been trying to build the ultimate supercomputer for a while now, but its no easy feat as Im sure you can imagine. There are currently three Department of Energy (DOE) Office of Science supercomputing user facilities: Californias National Energy Research Scientific Computing Center (NERSC), Tennessees Oak Ridge Leadership Computing Facility (OLCF), and Illinois Argonne Leadership Computing Facility (ALCF). All three of these supercomputers took years of planning and a lot of work to get them to the standard they are now, but its all been worth it as they provide researchers with the computing power needed to tackle some of the nations biggest issues.

There are two main challenges that supercomputers solve. The first is that it can analyze large amounts of data and the second is they can model very complex systems. Some of the machines about to go online are capable of producing more than 1 terabyte of data per second, which to put in laymans terms, is nearly enough to fill around 13,000 DVDs every minute. Supercomputers are far more efficient than conventional computers and calculations it could carry out in just one day would take 20 years for a conventional computer to calculate.

As mentioned earlier, the planning of a new supercomputer takes years and is often started before the last one has even finished being set up. Because technology moves so quickly, it works out cheaper to build a new one opposed to redesigning the existing one. In regards to the ALCF, staff began planning for it in 2008, but it wasnt until 2013 that it was launched. Planning involves not only deciding when and where it will be built and installed, but also deciding what capabilities the computers should have that is going to help with future research efforts.

When the OLCF began planning their current supercomputer, the project director, Buddy Bland, said, It was not obvious how we were going to get the kind of performance increase our users said they needed using the standard way we had been doing it. OLCF launched their supercomputer, Titan, in 2012 and combined CPUs (central processing units) with GPUs (graphics processing units). Using GPUs instead allows Titan to handle multiple instructions at once and run 10 times faster than OLCFs previous supercomputer. Its also five times more energy-efficient too.

Even getting the site ready to house the supercomputer takes time. When the NERSC installed their supercomputer, Cori, they had to lay new piping underneath the floor in which to connect the cooling system and cabinets. Theta is Argonnes latest supercomputer to go live which launched in July 2017.

There are many challenges that come with supercomputers too, unfortunately. One is that it literally has thousands of processors so programs have to break problems into smaller chunks and distribute them across the units. Another issue is designing programs that can manage failures. To help pave the way for future research, and to stress-test the computers also, in exchange for having to deal with this new computer issues, users are granted special access as well as being able to attend workshops and get hands-on help when needed.

Dungeon Sessions were held at NERSC while preparing for Cori. These were effectively three-day workshops, often in windowless rooms, where engineers would come together from Intel and Cray to improve their code. Some programs ran 10 times faster after these sessions. Whats so valuable is the knowledge and strategies not only to fix the bottlenecks we discovered when we were there but other problems that we find as we transfer the program to Cori, said Brian Friesen of NERSC.

But, even when the supercomputer is delivered its still a long way from being ready to work.First, the team that it goes to have to ensure that it meets all their performance requirements. Then, to stress-test it fully, they load it with the most demanding, complex programs and let it run for weeks on end. Susan Coghlan is ALCFs project director and she commented, Theres a lot of things that can go wrong, from the very mundane to the very esoteric. She knows this firsthand as when they launched Mira, they discovered that the water they had been using to cool the computer wasnt pure enough and as a result bacteria and particles were causing issues with the pipes.

Scaling up these applications is heroic. Its horrific and heroic, said Jeff Nichols, Oak Ridge National Laboratorys associate director for computing and computational sciences. Luckily the early users program gives exclusive access for several months before eventfully opening up to take requests from the wider scientific community. Whatever scientists can learn from these supercomputers will be used in the Office of Sciences next challenge, which is in the form of exascale computers computers that will be at least 50 times faster than any computer around today. Even though exascale computers arent expected to be ready until 2021, theyre being planned for now at the facilities and managers are already busy conjuring up just what they can achieve with them.

More News to Read

comments

Read the original:

How to Make a Supercomputer? - TrendinTech

Related Coverage

AUSTIN (KXAN) From his desk at an engineering building in Austin, University of Texas professor Clint Dawson has been tracking Hurricane Harvey powered by a supercomputer.

Dawson is part of a team of academics, from Louisiana to North Carolina, who have been trying to create better systems for compiling hurricane data since the 90s. Its a system Dawson says they can count on to work every time.

Using the strength of the Lonestar 5 supercomputer at the Texas Advanced Computing Center, Dawson and his colleagues download information from the National Hurricane Center and are able to update and automate it with high-resolution data. The information they gather are not forecasts but provide very detailed guidance which allows state agencies to see models for things like rising water levels with accuracy down to the neighborhood level.

As far as academic computing, this is the best available that we have to us in the country, Dawson said of the supercomputer they are using.

Dawson explained this is a level of detail that even National Hurricane Center forecasts dont achieve

He added that agencies like the Texas State Operations Center, TxDOT, NOAA, and the National Hurricane Center consult the data his team produces when making decisions like where to evacuate and where to send resources.

The State Operations Center can decide to use our guidance or they can decide to not use it, Dawson said. They tend to look at our results very carefully because we have done a good job of predicting hurricanes in the past.

Dawson is monitoring this data hourly, he says its exciting to provide information that is helpful to people on the ground.

This is what we live for, just like a storm chaser, like if you chase tornadoes you know you live for tornadoes. The flip side is people get hurt in these things and thats what were trying to do is prevent that, he said.

He added that he was surprised to note how much Harvey has escalated since Wednesday.

Yesterday it wasnt much and now today its something, you know its something big, he said. And its happening more and more, that hurricanes like this they blow up over night and suddenly we have a major event to deal with.

His team will be tracking the hurricane at least through Saturday, you can track their latest updates about the storm here. As of Thursday afternoon, their models anticipated a maximum water surge level of 12 feet in areas around where Harvey will make landfall.

More here:

UT Austin supercomputer helping agencies track Harvey - KXAN.com

The National Science Foundation (NSF) is soliciting proposals from US universities to acquire a $60 million next-generation supercomputer two to three times as powerful as Blue Waters.

The request for proposal (RFP) was originally published in May, and, as of July 14, all interested institutions were supposed to have sent the NSF a letter of intent, registering their interest. Final proposals are due on November 20. Its safe to assume that most, if not all of the major academic supercomputing centers in the US will be vying for the NSF grant.

The Pittsburgh Supercomputing Center (PSC), which collaborates with Carnegie Mellon University and the University of Pittsburgh, has gone on record about its intent to secure the funding for the Phase 1 system. An article published this week in the local Pittsburgh Post-Gazette reports that PSC would like to use such a machine to help spur the areas economy. Although the supercomputer would primarily be used by academic researchers in the science community, interim PSC director Nick Nystrom thinks the machine could also be a boon to the areas startup businesses, manufacturers and other industry players.

From the Post-Gazette report:

Everybody has big data, but big data has no value unless you can learn something from it, Mr. Nystrom said. We have a convergence in Pittsburgh: artificial intelligence, big data, health care and these are things PSC is already doing.

According to the Phase 1 RFP, the new system will be two to three times faster at running applications than the Blue Waters supercomputer, an NSF-funded machine installed at the National Center for Supercomputing Applications (NCSA), at the University of Illinois at Urbana-Champaign. Blue Waters is a Cray XE/XK system, powered by AMD "Interlagos" CPUs and NVIDIA K20X GPUs. It became operational in 2013.

Although Blue Waters has a peak speed of over 13 petaflops, NCSA never submitted a Linpack result for it. However, based on its peak performance, Blue Waters would almost certainly qualify as a top 10 system on the current TOP500 list. NCSA says a number of applications are able run at a sustained speed of more than one petaflop, with a plasma physics code attaining 2.2 petaflops. Given that the Phase 1 machine is supposed to be at least twice as powerful as Blue Waters, it should provide its users a significant boost in application performance.

This Phase 1 effort is also supposed to include an extra $2 million that will go toward the design of the Phase 2 system, which will be funded separately. That system is expected to be 10 times as fast as the Phase 1 machine, upon which it will draw at least some its technology and architecture. No hard dates have been set for this project.

The Phase 1 winner is anticipated to be announced in the first half of 2018, with the system expected to go into production by the end of FY 2019.

Continued here:

Next Big Academic Supercomputer Set for Debut in 2018 - TOP500 News

At the Hot Chips conference this week, Intel lifted the curtain a little higheron Knights Mill, a Xeon Phi processor tweaked for machine learning applications.

As part of Intels multi-pronged approach to AI, Knights Mill represents the chipmakers first Xeon Phi offering aimed exclusively at the machine learning market, specifically for the training of deep neural networks. For the inferencing side of deep learning, Intel points to its Altera-based FPGA products, which are being used extensively by Microsoft in its Azure cloud (for both AI and network acceleration). Intel is also developing other machine learning products for training work, which will be derived from the Nervana technology the company acquired last year. These will include a Lake Crest coprocessor, and, further down the road, a standalone Knights Crest processor.

In the meantime, its will be up to Knights Mill to fill the gap between the current Knights Landing processor, a Xeon Phi chip designed for HPC work, and the future Nervana-based products. In this case, Knights Mill will inherit most of its design from Knights Landing, the most obvious modification being the amount of silicon devoted to lower precision math the kind best suited for crunching on neural networks.

Essentially, Knights Mill replaces the two large double precision /single precision floating point (64-bit/32-bit) ports on Knights Landings vector processing unit (VPU), with one smaller double precision port and four Vector Neural Network Instruction (VNNI) ports. The latter supports single precision floating point and mixed precision integers (16-bit input/32-bit output). As such, it looks to be Intels version of a tensor processing unit, which has its counterpart in the Tensor Cores on NVIDIAs new V100 GPU. That one, though, sticks with the more traditional 16/32-bit floating point math.

The end result is that compared to Knights Landing, Knights Mill will provide half the double precision floating point performance, twice the single precision floating point performance. With the added VNNI integer support in the VPU (256 ops/cycle), Intel is claiming Knights Mill will deliver up to four times the performance fordeep learning applications.

The use of integer units to beef up deep learning performance is somewhat unconventional, since most of theseapplications are used to employingfloating point math. Intel, however, maintains that floating point offers little advantage in regard to accuracy, and is significantly more computationally expensive. Whether thistradeoff pans outor not remains to be seen.

Knights Mill will also support 16 GB of MCDRAM, Intels version of on-package high bandwidth memory assembled in a 3D stack, as well as 6 channels of DDR4 memory. From the graphic they presented at Hot Chips (above), the design appears to support 72 cores, at least for this particular configuration. Give the 256 ops/cycle value for the VPU, that would mean Knights Mill could deliver more than 27 teraops of deep learning performance for say, a 1.5 GHz processor.

Well find out what actual performance can be delivered once Intel starts cranking out the chips. Knights Mill is scheduled for launch in Q4 of this year.

Follow this link:

Intel Spills Details on Knights Mill Processor | TOP500 ... - TOP500 News

August 22, 2017 Ken Strandberg

Professor Satoshi Matsuoka, of the Tokyo Institute of Technology (Tokyo Tech) researches and designs large scale supercomputers and similar infrastructures. More recently, he has worked on the convergence of Big Data, machine/deep learning, and AI with traditional HPC, as well as investigating the Post-Moore Technologies towards 2025.

He has designed supercomputers for years and has collaborated on projects involving basic elements for the current and more importantly future Exascale systems. I talked with him recently about his work with the Tsubame supercomputers at Tokyo Tech. This is the first in a two-part article. For background on the Tsubame 3 system we have an in-depth article here from earlier this year.

TNP: Your new Tsubame 3 supercomputer is quite a heterogeneous architecture of technologies. Have you always built heterogeneous machines?

Satoshi Matsuoka, professor of the High Performance Computing Systems Group, GSIC, at the Tokyo Institute of Technology, showing off the Tsubame 3.0 server node.

Matsuoka: Ive been building clusters for about 20 years nowfrom research to production clustersall in various generations, sizes, and forms. We built our first very large-scale production cluster for Tokyo Techs supercomputing center back in 2006. We called it Tsubame 1, and it beat the then fastest supercomputer in Japan, the Earth Simulator.

We built Tsubame 1 as a general-purpose cluster, instead of a dedicated, specialized system, as the Earth Simulator was. But, even as a cluster it beat the performance in various metrics of the Earth Simulator, including the top 500, for the first time in Japan. It instantly became the fastest supercomputer in the country, and held that position for the next 2 years.

I think we are the pioneer of heterogeneous computing. Tsubame 1 was a heterogenous cluster, because it had some of the earliest incarnations of accelerators. Not GPUs, but a more dedicated accelerator called Clearspeed. And, although they had a minor impact, they did help boost some application performance. From that experience, we realized that heterogeneous computing with acceleration was the way to go.

TNP: You seem to also be a pioneer in power efficiency with three wins on the Green 500 list. Congratulations. Can you elaborate a little about it?

Matsuoka: As we were designing Tsubame 1, it was very clear that, to hit the next target of performance for Tsubame 2, which we anticipated would come in 2010, we would also need to plan on reducing overall power. Weve been doing a lot of research in power-efficient computing. At that time, we had tested various methodologies for saving power while also hitting our performance targets. By 2008, we had tried using small, low-power processors in lab experiments. But, it was very clear that those types of methodologies would not work. To build a high-performance supercomputer that was very green, we needed some sort of a large accelerator chip to accompany the main processor, which is x86.

We knew that the accelerator would have to be a many core architecture chip, and GPUs were finally becoming usable as a programming device. So, in 2008, we worked with Nvidia to populate Tsubame 1 with 648 third-generation Tesla GPUs. And, we got very good results on many of our applications. So, in 2010, we built Tsubame 2 as a fully heterogenous supercomputer. This was the first peta-scale system in Japan. It became #1 in Japan and #4 in the world, proving the success of a heterogeneous architecture. But, it was also one of the greenest machines at #3 on the Green 500, and the top production machine on the Green 500. The leading two in 2010 were prototype machines. We won the Gordon Bell prize in 2011 for the configuration, and we received many other awards and accolades.

It was natural that when we were designing Tsubame 3, we would continue our heterogeneous computing and power efficiency efforts. So, Tsubame 3 is the second-generation, large-scale production heterogeneous machine at Tokyo Tech. It contains 540 nodes, each with four Nvidia Tesla P100 GPUs (2,160 total), two 14-core Intel Xeon Processor E5-2680 v4 (15,120 cores total), two dual-port Intel Omni-Path Architecture (Intel OPA) 100 Series host fabric adapters (2,160 ports total), and 2 TB of Intel SSD DC Product Family for NVMe storage devices, all in an HPE Apollo 8600 blade, which is smaller than a 1U server.

A lot of the enhancements that went into the machine are specifically to make it a more efficient machine as well as for high performance. The result is that Tsubame 3although at the time of measurement for the June 2017 lists we only ran on a small subset of the full configurationis #61 on the Top 500 and #1 on the Green 500 with 14.11 gigaflops/watt, an RMax of just under 2 petaflops, and a theoretical peak of over 3 petaflops. Tsubme 3 just became operational August 1, with its full 12.1 petaflops configuration, and we hope to have the scores for the full configuration for the November benchmark lists, including the Top500 and the Green500.

TNP: Tsubame 3 is not only a heterogeneous machine, but built with a novel interconnect architecture. Why did you choose the architecture in Tsubame 3?

Matsuoka : That is one area where Tsubame 3 is different, because it touches on the design principles of the machine. With Tsubame 2, many applications experienced bottlenecks, because they couldnt fully utilize all the interconnect capability in the node. As we were designing Tsubame 3, we took a different approach. Obviously, we were planning on a 100-gigabit inter-node interconnect, but we also needed to think beyond just speed considerations and beyond just the node-to-node interconnect. We needed massive interconnect capability, considering we had six very high-performance processors that supported a wide range of workloads, from traditional HPC simulation to big data analytics and artificial intelligence, all potentially running as co-located workloads.

For the network, we learned from the Earth simulator back in 2002 that to maintain application efficiency, we needed to sustain a good ratio between memory bandwidth and injection bandwidth. For the Earth Simulator, that ratio was about 20:1. So, over the years, Ive tried to maintain a similar ratio in the clusters weve built, or set 20:1 as a goal if it was not possible to reach it. Of course, we also needed to have high bisection bandwidth for many workloads.

Todays processors, both the CPUs and GPUs, have significantly accelerated FLOPS and memory bandwidth. For Tsubame 3, we were anticipating certain metrics of memory bandwidth in our new GPU, plus, the four GPUs were connected in their own network. So, we required a network that would have a significant injection bandwidth. Our solution was to use multiple interconnect rails. We wanted at least one 100 gigabit injection port per GPU, if not more.

For high PCIe throughput, instead of running everything through the processors, we decided to go with a direct-attached architecture using PCIe switches between the GPUs, CPUs, and Intel OPA host adapters. So, we have full PCIe bandwidth between all devices in the node. Then, the GPUs have their own interconnect between themselves. Thats three different interconnects within a single node.

If you look at the bandwidth of these links, theyre not all that different. Intel OPA is 100 gigabits/s, or 12.5 GB/s. PCIe is 16 GB/sec. NVLink is 20 GB/sec. So, theres less than a 2:1 difference between the bandwidth of these links. As much as possible we are fully switched within the node, so we have full bandwidth point to point across interconnected components. That means that under normal circumstances, any two components within the system, be it processor, GPU, or storage, are fully connected at a minimum of 12.5 GB/sec. We believe that this will serve our Tsubame 2 workloads very well and support new, emerging applications in artificial intelligence and other big data analytics.

TNP: Why did you go with the Intel Omni-Path fabric?

Matsuoka: As I mentioned, we always focus on power as well as performance. With a very extensive fabric and a high number of ports and optical cables, power was a key consideration. We worked with our vendor, HPE, to run many tests. The Intel OPA host fabric adapter proved to run at lower power compared to InfiniBand. But, as important, if not more important, was thermal stability. In Tsubame 2, we experienced some issues around interconnect instability over its long operational period. Tsubame 3 nodes are very dense with a lot of high-power devices, so we wanted to be make sure we had a very stable system.

A third consideration was Intel OPAs adaptive routing capability. Weve run some of our own limited-scale tests. And, although we havent tested it extensively at scale, we saw results from the University of Tokyos very large Oakforest-PACS machine with Intel OPA. Those indicate that the adaptive routing of OPA works very, very well. And, this is critically important, because one of our biggest pain points of Tsubame 2 was the lack of proper adaptive routing, especially when dealing with degenerative effects of optical cable aging. Over time AOCs die, and there is some delay between detecting a bad cable and replacing it or deprecating it. We anticipated Intel OPA, with its end-to-end adaptive routing, would help us a lot. So, all of these effects combined gave the edge to Intel OPA. It wasnt just the speed. There were many more salient points by which we chose the Intel fabric.

TNP: With this multi-interconnect architecture, will you have to do a lot of software optimization for the different interconnects?

Matsuoka: In an ideal world, we will have only one interconnect, everything will be switched, and all the protocols will be hidden underneath an existing software stack. But, this machine is very new, and the fact that we have three different interconnects is reflecting the reality within the system. Currently, except for very few cases, there is no comprehensive catchall software stack to allow all of these to be exploited at the same time. There are some limited cases where this is covered, but not for everything. So, we do need the software to exploit all the capability of the network,including turning on and configuring someappropriate DMA engines, or some pass through, because with Intel OPA you need some CPUs involvement for portions of the processing.

So, getting everything to work in sync to allow for this all-to-all connectivity will require some work. Thats the nature of the research portion of our work on Tsubame 3. But, we are also collaborating with people like a team at The Ohio State University.

We have to work with some algorithms to deal with this connectivity, because it goes both horizontally and vertically. The algorithms have to adapt. We do have several ongoing works, but we need to generalize this to be able to exploit the characteristics of both horizontal and vertical communications between the nodes and the memory hierarchy. So far, its very promising. Even out of the box, we think the machine will work very well. But as we enhance our software portion of the capabilities, we believe the efficiency of the machine will become higher as we go along.

In the next article in this two-part series later this week, Professor Matsuoka talks about co-located workloads on Tsubame 3.

Ken Strandberg is a technical story teller. He writes articles, white papers, seminars, web-based training, video and animation scripts, and technical marketing and interactive collateral for emerging technology companies, Fortune 100 enterprises, and multi-national corporations. Mr. Strandbergs technology areas include Software, HPC, Industrial Technologies, Design Automation, Networking, Medical Technologies, Semiconductor, and Telecom. He can be reached at ken@catlowcommunications.com.

Categories: HPC

Tags: OmniPath, Tsubame

An Early Look at Baidus Custom AI and Analytics Processor First In-Depth View of Wave Computings DPU Architecture, Systems

View post:

Inside View: Tokyo Tech's Massive Tsubame 3 Supercomputer - The Next Platform

Pittsburgh Post-Gazette

Pittsburgh stepping up to try to win competition for supercomputer project Pittsburgh Post-Gazette Pittsburgh is competing to build the fastest nongovernmental computer in the country, with an economic impact to the region that could run up some big numbers possibly exceeding $1 billion, according to one backer. It would also need a lot of power ...

Continue reading here:

Pittsburgh stepping up to try to win competition for supercomputer project - Pittsburgh Post-Gazette

China is planning to boost its computing power tenfold within a couple of years by building a new generation supercomputer.

The machine will be based on the coast of Shandong province to process the data collected from the worlds oceans, according to scientists briefed on the project.

An Hong, professor of computer science with the University of Science and Technology of China in Hefei and a member of a committee advising the central government on high performance computer development, said the worlds first exascale computer would have a dedicated mission of helping Chinas maritime expansion.

An exascale computer is defined as one that can carry out one billion billion calculations per second. It is not only 10 times faster than Sunway Taihulight at present the worlds fastest computer which operates from Wuxi, Jiangsu but equal to the calculation power of all the worlds top 500 super computers combined.

An said the machine could be finished as soon as 2019. Three independent supercomputer manufacturers on the mainland are competing for the contract. They include Sugon, or the Dawning Information Industry, which is owned by the Chinese Academy of Sciences; the National University of Defence Technology, which built the Tianhe series supercomputers, and the Sunway team. They have produced blueprints featuring vastly different architectures, according to An.

The authorities are looking to pick a design that not only offers a high performance but will be ready for immediate use once built. The budget for the project is expected to be between one and two billion yuan (US$150 million-US$300 million).

Chinese scientists create biggest virtual universe with worlds fastest computer, beating European record

The most important question to us is not whether China can build an exascale computer, or how fast, but why, An said.

There is indeed a race among nations on supercomputers, but this is not our concern. Our concern is the ocean, she added.

When elected leader of the Chinese Communist Party in 2012, President Xi Jinping promised to turn China into a hai shang qiang guo, or maritime superpower, with an expansionist policy that would, according to the state media, be comparable with the fleet of Zheng He established during Ming dynasty six centuries ago to spread Chinas influence over the world.

Soviet-built Liaoning pales in comparison to Zheng Hes majestic fleet

Within the space of a few years, China has effectively tightened its grip on the South China Sea, dismissed numerous neighbours claims over disputed waters, acquired military ports in South Asia and the African east coast, developed some of the worlds most advanced nuclear submarines with electromagnetic drive, explored vast areas of the sea bed for energy and mineral deposits, and launched the Belt and Road Initiative to strengthen economic ties with other countries, the belt roughly following Zheng Hes ports of call.

Chinese vessels, naval outposts and unmanned monitoring facilities including a global network of buoys, satellites, sea floor sensors and underwater gliders are generating countless steams of data every second.

According to marine researchers, these data contain a rich variety of information such as sea current readings, trace chemicals, regional weather and anomalies in water density that could be used for anything from helping submarines avoid turbulence to negotiating cuts to green house gas emissions.

Why Beijing is speeding up underwater drone tests in the South China Sea

Feng Liqiang, operational director of the Marine Science Data Centre in Qingdao, Shandong said the exascale computer would be able to pull all marine-related data sets together to perform the most comprehensive analysis ever.

It will help, for instance, the simulation of the oceans on our planet with unprecedented resolution. The higher the resolution, the more reliable the forecast on important issues such as El Nino and climate change, he said.

It will give China a bigger say over international affairs, Feng added.

In June, the US government commissioned six companies including IBM, Intel and Hewlett Packard Enterprise to come up with countermeasures against Chinas lead on high performance computing.

China not only hosts Sunway and Tianhe, which currently rank first and second on international performance charts with speeds that far exceed those of their foreign competitors, but also overtook the US last year in terms of installed supercomputing capacity, an event described as an inflection point by Horst Simon, deputy director of the Lawrence Berkeley National Laboratory in California.

The White House currently hopes that American companies will be able to come up with a design to have an exascale computer up and running by 2021.

China hits milestone in developing quantum computer to eclipse all others

A major challenge for exascale computing is cooling. Researchers have struggled to find a way to reduce the tremendous heat generated by a large number of chips. The problem not only prevents the computer from reaching peak performance but also leads to large electricity bills.

Though the exact location of the Chinese exascale computer has not yet been determined by the authorities, Zhang Haichun, professor at the computer science and technology department, the Ocean University of China in Qingdao, said the city, the largest port in Shandong, had numerous advantages as a home for the exascale computer.

Qingdao has more ocean-related research institutes than any other city along Chinas coastline. The worlds largest marine data centre is under construction, and it is directly linked to the nations major monitoring networks above and under the ocean, he said.

Putting the machine in Qingdao will save the trouble of transmitting a large amount of data over long distance through optic fibres. An expensive data plan can break the projects bank account, Zhang said.

But marine scientists said that a single computer would not bridge the maritime power gap between China and the US overnight.

Our data is increasing at a fast pace but it is still dwarfed by the amount gathered by the US through decades of patient, continuous effort, said Feng.

We also lack the powerful software with sophisticated algorithms to handle the data at full exascale speed, he added.

Lu Xianqing, professor at the Key Laboratory of Physical Oceanography under the Ministry of Education in Qingdao, said he had serious doubts whether the project would fulfil its mission due to the difficulty of accessing marine-related data in China.

Why China still cant beat US to become the worlds most powerful navy

Unlike in the US, where most ocean data gathered from public-funded research is open to access after a limited protection period, there is no official information sharing mechanism on the mainland.

The State Ocean Administration runs and hoards its own data sets, as do the PLA Navy, the Chinese Academy of Sciences and many universities. Every institute is treating data as private asset for the interests of their own research he said.

There are also fears that the exchange of data might lead to leaks and threaten national security. Foreign submarines, for instance, need detailed information on Chinas offshore areas. These security concerns have prompted government officials to classify most data collected from the sea as confidential.

Getting data for the computer may turn out to be more difficult than building it, Lu said.

An said the Chinese government might approve the construction of another exascale computer, but its purpose and location was still uncertain uncertainty.

Besides US and China, the European Union and Japan are also trying to build an exascale computer by around 2020.

See the rest here:

The world's next fastest supercomputer will help boost China's ... - South China Morning Post

eResearch South Australia (eRSA) has boosted its services to the state's research, government, and business sectors, announcing the arrival of its new high-performance computer (HPC) and research cloud, Tango.

Built on Dell EMC infrastructure, and the cloud on VMware Cloud Foundation, Tango is expected to provide South Australia's researchers with a platform to accelerate innovation, eRSA infrastructure manager Paul Bartczak told ZDNet.

As eRSA is a research-focused joint venture between the University of Adelaide, Flinders University, and the University of South Australia, Bartczak said big data and complex data analytics are increasingly becoming a part of the research landscape.

Founded in 2007, the not-for-profit incorporated entity provides "not normal IT" services to researchers and commercial users in the state, including high-performance computing, cloud computing, big data storage, management and analysis, software development, and consultancy.

"In order to keep up with modern research demands, having a powerful HPC and cloud platform is a necessity," Bartczak explained. "Tango allows researchers and commercial clients alike to process large amounts of data quickly and easily, and the software-defined network capability provides a secure environment to work in. Tango is highly scalable, which means it can scale easily to keep up with increased user demand."

Bartczak expects Tango will provide its users with freedom in the Research and Innovation Sandbox, which is a service allowing users to experiment with different combinations of technology and test the best ways of processing analytics, big data, complex modelling, and forecasting.

"People come to us and ask what makes us different from the likes of Amazon or Azure or other traditional IT service providers -- we're here to support and foster that innovation, development, and research space, and provide that environment for people [to experiment]," he explained.

"The cost of trial and error can be quite significant, but we like to see ourselves as enablers in that area, giving people this environment to try, fail, keep developing, and, once they require production-grade systems, then they can move onto a traditional service provider to implement things."

eRSA caters for a range of groups furthering research on terrestrial ecosystems, ancient DNA, physics, chemistry, genomics, biology, arts, humanities, and everything in between, Bartczak said, adding that the organisation boasts a large user base.

"We're here to basically create solutions -- whether it's cloud or HPC-related -- for any discipline," he added.

Tango comprises Dell PE R730 server with Intel Xeon Processor E5-2690v4 35M, connected with Dell Z9100-ON 100GbE software-defined network switches.

The HPC packs 32GB RDIMM 2400MT/s DDR4 SDRAM; 200GB SSD 6Gbps; and Mellanox ConnectX-4 Dual Port 25GbE DA/SFP.

The cloud boasts Dell PE R730xd Servers connected with Intel Xeon Processor E5-2680 v4 35M; 32GB RDIMM 2400MT/s DDR4 SDRAM; 1.6TB SSD 6Gbps and 6TB 7.2K RPM NLSAS; and Mellanox ConnectX-4 Dual Port 25GbE DA/SFP.

According to Andrew Underwood, Dell EMC's Australia and New Zealand HPC lead, Tango has been built to require minimal touch from the eRSA teams to allow them to focus on the scientific outcomes of the state's researchers instead of keeping the lights on.

"One of the really nice things that I like about what our partnership with eRSA is this system is really going to be helping power the South Australian ecosystem of research and innovation, as well as startups," Underwood told ZDNet.

"One of the things that eRSA has really looked at, and been quite visionary in, is that small to medium enterprises (SMEs) have not been so fortunate when it comes to accessing high-performance computing in the past -- they typically have tight budgets and limited people resources, so the new Tango system ... will mean that SMEs who are either wanting to access the system themselves or partnering with one of the universities in South Australia will be able to access this."

Tango will replace eRSA's previous HPC Tizard that will be decommissioned on August 31.

Last week, Dell EMC announced that it will be providing Melbourne's Swinburne University of Technology with an HPC to power research into astrophysics and gravitational waves, with the university seeking to further prove the science behind Einstein's theory of general relativity.

OzSTAR, which loosely stands for the Australian supercomputer for theoretical astronomical research, will be built by the tech giant at a cost of AU$4 million, and will be used by the Swinburne-led Australian Research Council Centre of Excellence for Gravitational Wave Discovery.

Read more:

It takes two to Tango: eRSA partners with Dell EMC for new research supercomputer - ZDNet

Melbournes Swinburne University is going to deploy its first petascale supercomputer, a Dell EMC machine that will be tasked to support cutting-edge astrophysics and other scientific research.

The $4 million supercomputer, known as OzStar, will be comprised of 115 PowerEdge R740 nodes, each of which will be equipped with two of Intels Xeon Skylake Scalable processors and two NVIDIA Tesla P100 GPUs. An additional four nodes are to be powered by Intel Xeon Phi processors. The nodes will be connected with the Intel Omni-Path fabric, operating at 100Gbps. Peak performance is expected to be in excess of 1.2 petaflops. The system will also be equipped with five petabytes of Lustre storage, comprised of Dell EMC PowerVault MD3060e enclosures.

OzStar will eclipse the existing Green II systems (gSTAR and SwinSTAR), SGI machines installed at Swinburne in 2011 and 2012, which are accelerated by the now-ancient NVIDIA Tesla C2070 and K10 GPUs. The even-older Green machine, which is still in operation, is a Dell PoweEdge cluster deployed in 2007. That system is powered by Intel Xeon Clovertown processors.

According to Chris Kelly, Dell EMC VP and GM for the Compute and Networking group covering Australia and New Zealand, the new OzStar system will spend around a third of its time processing gravitational wave data collected by advanced LIGO (Laser Interferometer Gravitational-wave Observatory) installations in the US. The LIGO detectors are able measure ever-so-small ripples in space-time caused large-scale cosmic events, such as the collision of black holes and neutron stars, and the explosions of supernovae.

Sifting through this LIGO data has enabled researchers to detect the locations of these objects in the far reaches of the universe and study their behavior. However, such data analysis requires enormous amounts of computation, and thats why Swinburnes new petascale supercomputer is expected to be an important addition to gravitational-wave astrophysics, a research domain that began in earnest with the detection of the first such waves in 2015. Its exciting to think [OzStar] will be making advances in a field of study that didnt really exist two years ago, wrote Kelly.

The system is currently being installed at Swinburnes Centre for Astrophysics & Supercomputing, which is the headquarters of the Centre of Excellence for Gravitational Wave Discovery, also known as OzGrav. The new center, which opened for business earlier this year, was set up by the Australian Research Council (ARC), with Swinburne specifically, the Swinburne University of Technology as the lead institution. However, access to OzStar will be available to astrophysicists throughout Australia.

Beside gravitational wave studies, the system will also be used to support research in molecular dynamics, nanophotonics, advanced chemistry and atomic optics. OzStar is scheduled to be up and running by the end of August and available for full production in September.

Continue reading here:

Swinburne University Makes Leap to Petascale with New Supercomputer - TOP500 News

On Monday, at 12:31 p.m. Eastern time, a SpaceX Falcon 9 rocket lifted off on a resupply flight for the International Space Station, and among its cargo, in addition to ice cream, was something else very cool: a supercomputer.

The machine, made by Hewlett Packard Enterprise and called the Spaceborne Computer, is capable of a teraflop worth of computing power, which puts it roughly in line with a late-1990s supercomputer. Made up of two pizza box-shaped machines in a single enclosure, the HPE supercomputer is a part of a year-long experiment to see how an off-the-shelf computer system can fare in space if protected in the right way by software.

Long space missions like a trip to Mars come with considerable communications delays, so equipping astronauts with a powerful supercomputer would allow them to solve complex problems without having to wait for the issue and the solution to be transmitted to and from Earth. But radiation on a trip like that can damage computers, so NASA and HPE are conducting this research to see if software can provide the necessary protection to keep things functioning correctly.

Just like NASAs famous identical twin experimentin which Scott Kelly spent a year in space and his brother, Mark Kelly, stayed down on Earththe supercomputer in space has a brother on this planet, a doppleganger machine located in Wisconsin acting as a control in this experiment.

HPEs approach with the Spaceborne Computer, a two-node, water-cooled machine, is different from the way a mission-critical computer in space is physically protected (or hardened, in space gear speak) from radiation. For example, the chief computer for the Juno spacecraft inhabits a protective titanium vault with walls about one centimeter thick, according to BAE systems, which made that processor. Instead of physical protection for the HPE computer, the company is hoping to learn if software can do something similar.

Eng Lim Goh, the HPE projects principal investigator, says that the dramatic vision for the future of this line of research is one in which before an astronaut travels to space, he or she would be able to take a top-of-the-line, off-the-shelf machine with them, and software could make it space-worthy. Then the astronaut could put whatever programs she wanted on the machine, a process that Goh, a computer scientist, compares to having an iPhone in space onto which youve preloaded your apps.

So how might this computer's software help protect it?

In general, Goh says that smart machines on Earth that exercise self-care may turn themselves off in the face of dangerous conditions. Another idea is that a machine can intentionally run slowly so that it can handle errors as it goes, as opposed to running at maximum capacity and not having the bandwidth to also cope with problems.

We will find out what works, what doesnt, Goh says. We have a whole list.

HPE said in a statement describing the project that the system software will manage real time throttling of the computer systems based on current conditions and can mitigate environmentally induced errors. (If youre wondering what Hewlett Packard Enterprise is, its one-half of the old HP, which divided in 2015; the other half is now HP, Inc., which makes personal computers and printers.)

This system is not planned to replace the [physically] hardened systems, Goh says. The intention is that something like this could function as a decision support tool on a long mission to a place like Mars, and not a primary mission computer.

Its due to arrive at the space station in the Dragon Spacecraft on Wednesday.

This article has been updated to correct errors with the spelling of Eng Lim Goh's name.

View original post here:

NASA is about to find out if a supercomputer can survive a year in space - Popular Science

Upcoming additions to MareNostrum 4 will bring its total speed up to 13.7 petaflops.

Barcelona's Supercomputing Center is aiming to lead the move in Europe from petascale to exascale computing -- one exaflops is 1,000 petaflops -- and the newly launched MareNostrum 4 is just part of that shift.

The 34m ($40m) MareNostrum 4, which recently began operations, is the third fastest supercomputer in Europe and occupies 13th place in the Top500 list of the world's high-performance computing systems.

It provides 11.1 petaflops for scientific research, 10 times more than MareNostrum 3, which was installed between 2012 and 2013. One petaflops is one thousand million million floating-point operations per second.

That performance means the new supercomputer will be able to deal more quickly with tasks relating to climate change, gravitational waves, fusion energy, AIDS vaccines, and new radiotherapy treatments to fight cancer.

On top of that power, the capacity of the general-purpose cluster is also due to be upgraded in the next few months with the addition of three new smaller-scale clusters. The general-purpose cluster is the largest and most powerful part of the supercomputer, consisting of 48 racks with 3,456 nodes and 155,000 processors.

The new clusters will be equipped with emerging technologies developed in the US and Japan, such as IBM Power9 and NVIDIA Volta Plus GPUs, Intel Knights Hill processors or 64-bit ARMv8 processors, used in powerful supercomputers such as American Summit, Sierra, Theta and Aurora, and Japanese Post-K.

Those additions will bring MareNostrum 4's total speed up to 13.7 petaflops. For comparison, the system occupying seventh place in the Top500, Japan's Joint Center for Advanced High Performance Computing Oakforest-PACS, offers 13.5 petaflops.

MareNostrum 4's storage capacity will also increase to 14 petabytes. However, despite those processing and storage additions, its energy consumption will only increase by 30 percent to 1.3MW per year.

"For our researchers in computer architecture, and in the programming and creation of tools for the analysis and efficiency of computers, this a treat," BSC director professor Mateo Valero tells ZDNet.

"It will allow us to experiment with cutting-edge technologies, analyze how the same applications behave in different hardware, and tackle the challenge of making them efficient in different architectures.

Valero said the upcoming changes will also enable BSC to test the suitability of these technological developments for future iterations of MareNostrum.

"It will also allow us to address one of our most ambitious projects: our participation in the creation of hardware and software technology with a European DNA," he says.

For the first time, the European Commission is supporting that goal. Last March, seven European countries including Spain signed a formal declaration to support Europe's leadership in high-performance computing, a project of the size of Airbus in the 1990s and of Galileo in the 2000s.

At the time of the declaration, Andrus Ansip, European Commission vice-president for the digital single market, said that if Europe stays dependent on others for this critical resource, then it risks getting technologically "locked, delayed, or deprived of strategic know-how".

'Technological sovereignty' is about being technologically independent, so that you have control of your research and development, which can be particularly important for national security.

BSC plans to capitalize on the achievements and knowledge of its researchers to create European processors for supercomputing, the automotive industry, and the Internet of Things.

Of course, the transition to exascale computing requires an implementation roadmap. According to Valero., BSC doesn't expect to have an exaflops machine in 2020, but Europe could and should have one in 2022 or 2023.

How supercomputing power is helping with anti-pollution plans like city-wide car bans

A modeling tool developed at Barcelona's Supercomputing Center is busy predicting levels of atmospheric pollutants in Spain, Europe, and now Mexico.

Using ARM chips and Linux, Barcelona center dreams of being 'Airbus of supercomputing'

A chapel in the heart of Barcelona University is home to one of Europe's most powerful supercomputers - and a mobile chip-based successor is under development.

See the original post here:

MareNostrum 4 supercomputer now runs at 11 petaflops but there's more to come - ZDNet

CHARLIE MITCHELL

Last updated14:52, August 18 2017

JOSEPH JOHNSON/STUFF

Albi Whale and his father, Dr David Whale, left.

It runs an internationalcompany, helps manage a doctor's office on the side and soon "Zach" will be the face of artificial intelligence (AI) in Christchurch.

Zach is billed as one of the world's most powerful supercomputers, an AI system that interacts with people like they do each other.

It is expected to be on display in a restored heritage building inChristchurch by 2019, withan education centre and virtual classrooms, and ways for the public to have conversations with it.

The non-commercial technologywas bought and adapted by the Terrible Foundation, a social enterprise run by Christchurch-based entrepreneur Albi Whale.

READ MORE: *AI learns from pro gamers - then crushes them *Technology whizz-kid says the future looks bright, not evil *How artificial intelligence is taking on ransomware *Artificial intelligence has mastered painting

Whale founded Terrible Talk, a non-profit internet and phone provider.

Earlier this year, Zach became chief executive of Terrible Talk:Itruns virtually the entire company, including handling the company's accounts, making management decisions, andanswering customer queries via email.

Whale's father and colleague, Dr David Whale, said Zach was unlike other AI systemsin that it was built from the ground up around human interaction.

"You can talk to it, write to it. You can draw pictures andit will respond.This is a system that interacts with us the sameway we interact with each other."

One of Zach's most promising applicationswasin the healthcare system, as a digital assistant.

For the last six weeks, Christchurch GP Dr Rob Seddon-Smith has used it to handle tasks in his Hei Hei clinic.

Seddon-Smith who has been teaching the AI, which improves itselfthrough feedback presented his findings on Thursday. He said they were astonishing.

The AI listens to his consultations and writes up the patient notes. It doesn't transcribe, but truncates and expresses the important parts of the conversation ina readable way they were vastly better than Seddon-Smith's own notes, he said.

"He can listen to the consultation, capture the very essence of the words and record them in a recognisable form. It works," he said.

"This set of notes is the first ever, anywhere in the world, to be created only by computer. I didn't type anything, I simply chatted with my patient."

Other AI, such as Apple's Siri,"couldn't do anything close" to what Zach could, he said.

Patients would be able to ring and askit for their medical information,make appointmentsand have questions answered. It recognises voice patterns to verify identities.

Tests attempting to break its security systemshad been unsuccessful, including by its own creators.

What clinched Seddon-Smith's belief in Zach's capabilities was when, unprompted, it put the phone number for a crisis hotline into its notes for a suicidal patient.

It textedhim one night despite not having his phone number to tell him his email inbox was full.

Ittook away all the mundane tasks doctors had to doand allowed him to focus on his patients.

"It can address some of the most complex issues in healthcare and do so efficiently, safely and above all, equitably. It is one technology built from the ground-up to leave no one behind."

Councillor Deon Swiggs said it was expected the AI would be installed in a restored heritage building, mixingthe city's past with its future.

"It's exciting that by 2019, Christchurch will be home to one of the world's largest supercomputers.It's actually reallyincredibleto think about," he said.

"The investment here is huge, and I don't think that can be understated. It will stimulate tech tourism, a massive industry . . . it will increase Christchurch's credentials as a city of opportunity and of technology."

There were lots of questions about the impact AI would have in the future, particularly for people'sjobs, he said.

"I think it's really important to have an AI in Christchurch that we are going to be able to integrate with and engage with, so people can take away the fear of what these things are."

In its current form,Zachcan speak and holdconversations, but its voice capacity is turned off as it is too resource intensive.

By the time it isinstalled in Christchurch, it is expected to have greater capacity, and will be able to hold conversations with the public.

-Stuff

Read more:

Zach, a supercomputer that can hold conversations, is coming to Christchurch - The Press

Dell EMC announced yesterday it is building a new supercomputer the OzStar for the Swinburne University of Technology (Australia) in support theARC Centre of Excellence for Gravitational Wave Discovery (OzGrav). The OzGrav project was first announced last September. The OzStar supercomputer will be based on Dell EMC PowerEdge R740 nodes, have more than one petaflops capability, and is expected to be completed in September.

OzGrav will use the new machine in support of efforts to understand the extreme physics of black holes and warped space-time. Among other projects, OzGrav will process data from LIGO (Laser Interferometer Gravitational Wave Observatory) gravitational wave detectors and the Square Kilometre Array (SKA) radio telescope project with facilities built in Australia and South Africa.

The OzStars architecture will leverage advanced Intel (Xeon V5) and Nvidia (P100) technology and feature three building blocks: Dell EMC 14thGeneration PowerEdge R740 Servers; Dell EMC H-Series Networking Fabric; and Dell EMC HPC Storage with Intel Lustre filesystem. The networking fabric is Intel Omni Path Architecture and will provide 86.4 Terabits per second of aggregate network bandwidth at 0.9 s latency according to Dell EMC. As is typical in such contracts, Dell EMC will provide support.

Heres snapshot of OzStars specs as provided by Dell EMC:

While Einstein predicted the existence of gravitational waves, it took one hundred years for technology to advance to the point they could be detected, said Professor Matthew Bailes, director of OzGrav, Swinburne University of Technology. Discoveries this significant dont occur every day and we have now opened a new window on the Universe. This machine will be a tremendous boost to our brand-new field of science and will be used by astrophysicists at our partner nodes as well as internationally.

This combination of Dell EMC technologies will deliver the incredibly high computing power required to move and analyze data sets that are literally astronomical in size, said Andrew Underwood, Dell EMCs ANZ high performance computing lead, who collaborated with Swinburne on the supercomputer design.

The NSF-funded LIGO project first successfully detected gravitational waves in 2015. Those waves were caused by the collision of two modest size black holes spiraling into one another (see HPCwire article,Gravitational Waves Detected! Historic LIGO Success Strikes Chord with Larry Smarr). LIGO has since detected two more events opening up a whole new way to examine the universe.

According to todays announcement, up to 35% of the supercomputers time will be spent on OzGrav research related to gravitational waves. The supercomputer will also continue to incorporate the GPU Supercomputer for Theoretical Astrophysics Research (gSTAR), operating as a national facility for the astronomy community funded under the federal National Collaborative Research Infrastructure Scheme (NCRIS) in cooperation with Astronomy Australia Limited (AAL). In addition, the supercomputer will underpin the research goals of Swinburne staff and students across multiple disciplines, including molecular dynamics, nanophotonics, advanced chemistry and atomic optics.

OzStar replaces the green machines that have served Swinburne for the last decade and seeks to further reduce Swinburnes carbon footprint by minimizing CO2 emissions by carefully considering heating, cooling and a very high performance per watt ration of power consumption.

OzGrav is funded by the Australian Government through the Australian Research Council Centres of Excellence funding scheme and is a partnership between Swinburne University (host of OzGrav headquarters), the Australian National University, Monash University, University of Adelaide, University of Melbourne, and University of Western Australia, along with other collaborating organisations in Australia and overseas.

See original here:

Dell EMC will Build OzStar Swinburne's New Supercomputer to Study Gravity - HPCwire (blog)

Microsoft has bought Cycle Computing, an established provider of cloud orchestration tools for high performance computing users. The acquisition offers the prospect of tighter integration between Microsoft Azures infrastructure and Cycles software, but suggests an uncertain future for the technology on Amazon Web Services (AWS) and Googles cloud platform.

In a blog posted by Cycle Computing CEO and co-founder Jason Stowe, he noted the advantages of Azures scale and market position, writing:

Its global cloud footprint and unique hybrid offering is built with enterprises in mind, and its Big Compute/HPC team has already delivered pivotal technologies such as InfiniBand and next generation GPUs. The Cycle team cant wait to combine CycleClouds technology for managing Linux and Windows compute & data workloads, with Microsoft Azures Big Compute infrastructure roadmap and global market reach.

Cycle Computing was founded by in 2005, without the benefit of venture capital or large investors. According to Stowe, the company was bootstrapped using $8,000 charged against a credit card. That was apparently enough to launch CycleCloud, a software suite that provides cluster provisioning, configuration, monitoring, and optimization. Today, CycleCloud is used to manage a billion core-hours of cloud computing, primarily on infrastructure provided by AWS, Google, and Microsoft Azure, as well as on in-house clusters. Stowe claims the business, measured in core-hours, is growing 2.7x per year.

CycleCloud is applicable to essentially any HPC domain, but it most prevalent in the areas of biotech/pharma, manufacturing, financial services, digital content creation, and scientific research. Cycle made a name for itself by orchestrating some of the largest on-demand clusters ever attempted. A record run on 156,000 cores, using EC2 spot instances with AWS, was used to evaluate 220,000 candidate compounds for building better solar cells. That work was done by researchers at the University of Southern California (USC). Besides UCS, other marquee customers with big cloud jobs are Novartis, Pacific Life, Johnson & Johnson, HGST, and the Broad Institute.

Microsoft, of course, is no stranger to big customers with equally big demands. Azure already supports HPC-type workloads, but not in the shrink-wrapped way that the Cycle platform does. In fact, CycleCloud uses the low-level API hooks on cloud platforms like Azure to provision the hardware and configure the software stack for the user. It also monitors the executing job to dynamically optimize allocation and use of resources. Any of this can be accomplished on bare Azure; it just requires more programming effort.

Since CycleCloud already supports Azure, what does Microsoft get out of this? In a separate blog penned by Jason Zander, Corporate Vice President, Microsoft Azure, he says the combo of the Azure infrastructure and the Cycle technology will open up many new possibilities.

One of those possibilities is better Linux support. It should be noted that Azure already supports Linux containers, but apparently Zander thinks Cycle can help further this capability. Its no mystery why Microsoft would consider this important, inasmuch as Linux applications are taking a growing share of Azure jobs. A ZDNet article from last year reported that Microsoft has gone from one in four of its Azure virtual machines running Linux to nearly one in three. For its part, CycleCloud supports both Linux and Windows, but the vast majority of HPC customers are devoted to Linux, something Microsoft came to grips with several years ago when it was peddling its Windows HPC Server OS.

Another potential advantage to bringing Cycle in-house is offering better support for some of the fastest growing application domains in the cloud, namely artificial intelligence, the internet of things, and deep learning. Here Zander points to Cycle Computings depth and expertise in massively scalable applications. Especially in the AI arena, CycleCloud could provide a convenient orchestration of Azures GPU instances and FPGA-powered servers. Combined that with Azures Cognitive Services, and you have the makings of a formidable AI platform.

While all of that paints an interesting portrait of CycleCloud and Azure, current customers on AWS and Google will likely find themselves left out of the picture. When asked to provide some detail on what will become of the Cycle support for other cloud providers, Microsoft offered this:

We will continue to support Cycle Computing clients using AWS and/or Google Cloud. Future Microsoft versions released will be Azure focused. We are committed to providing customers a seamless migration experience to Azure if and when they choose to migrate.

Which is another way of saying if you want to remain a CycleCloud customer going forward, youre going to end up on Azure. Using the prospect of future Cycle support to encourage such migration could be an unstated motivation for buying the company. Microsoft is locked in a battle with both Google and Amazon for cloud computing dominance, with Amazon currently the clear leader. Any leverage that Microsoft can used to peel off customers from its rivals, especially customers needing massive amounts of servers, works to its advantage. And by that criteria alone, the Cycle Computing acquisition looks like a wise move.

See the original post:

Microsoft Boosts HPC Cloud Aspirations with Acquisition of Cycle Computing - TOP500 News

The International Space Station is waiting to welcome its very first supercomputer

HP Enterprise is looking to give the space computing a massive upgrade with its new supercomputer. Dubbed as the Spaceborne Computer, this supercomputer will be launching on a SpaceX Falcon 9 rocket to add some serious computing power to the International Space Station (ISS).

Using a computer on the ISS isnt as easy as on Earth. Unlike Earth, where you just need have power and a good internet connection to make your computer work, computers on the ISS need to be connected using satellite network, which isnt as good as the internet on Earth. Also, computers and laptops used in space have been really slow to reach such powerful data processing abilities. A lot of computers and hardware have now undergone specific retrofitting via a process called hardening for extra protection to survive the harsh conditions of space.

Image Credit: HP Enterprise

Currently it can take years to harden a computer, says Mark Fernandez, Americas Technology Officer at HPE and leading payload engineer for the project. By the time its finished its mission, it could be three to five generations old.

However, HPEs supercomputer is different. It gives up much of the physical ruggedizing for software, which will theoretically make up for the condition on the ISS. As its meant to be used for only a year, no hardening will be required.

This is a general-purpose high-performance computing Linux-based system, says Fernandez. All of the top 500 [supercomputers] run similar to this. Scientists will be able to to focus on science and not the networking the downlink internet from the ISS isnt that good.

If the Spaceborne Computer still functions perfectly after one year, it will be a milestone for NASA to send up even more powerful computers. And, in the near future, astronauts traveling to Mars could use the similar type of computers. The Falcon 9 rocket carrying the supercomputer will launch on Monday, August 14.

Continue reading here:

The International Space Station is waiting to welcome its very first supercomputer - The TechNews

Dell EMC has announced it will be providing Melbourne's Swinburne University of Technology with a high performance supercomputer to power research into astrophysics and gravitational waves, with the university seeking to further prove the science behind Einstein's theory of general relativity.

OzSTAR, which loosely stands for the Australian supercomputer for theoretical astrological research, will be built by Dell EMC at a cost of AU$4 million and will be used by the Swinburne-led Australian Research Council (ARC) Centre of Excellence for Gravitational Wave Discovery (OzGrav).

"While Einstein predicted the existence of gravitational waves, it took 100 years for technology to advance to the point they could be detected," said OzGrav director, professor Matthew Bailes.

"Discoveries this significant don't occur every day and we have now opened a new window on the universe. This machine will be a tremendous boost to our brand new field of science."

Speaking with ZDNet, Andrew Underwood, HPC lead at Dell EMC Australia and New Zealand, explained the supercomputer will be used to process incredibly large volumes of data coming via giant telescopes, including the $1 billion Square Kilometre Array (SKA), slated the largest and most capable radio telescope ever constructed.

Touted as the world's largest science project, involving 20 countries and covering over 1 million square metres of data collection area, the SKA has its central cores of operation in South Africa and Western Australia, with its central computer alone boasting the processing power of about 100 million PCs.

Swinburne will be connecting OzSTAR to research collaborators in the country's west, with the university also involved in a number of projects to help deliver the SKA, including designing the telescope's antennas, in addition to data processing.

OzSTAR is the fifth supercomputer since Swinburne received its first back in 2001. Only the first installed at the university was delivered by Dell. Underwood said reading about the supercomputer 16 years ago inspired him to join the tech giant in its supercomputing ventures.

OzGrav, which received AU$31.1 million in September to carry out its research, is also going to be focused on education in varying levels of schooling, hoping to bring youngsters into the world of science.

"Analysing the ripples in space-time to understand the birth of the universe doesn't get much cooler than that," Underwood added.

The supercomputer will comprise new Dell EMC 14th generation PowerEdge Servers, containing Intel Skylake Xeon processors and Nvidia Tesla P100s; Dell EMC H-Series networking fabric powered by Intel's Omni-Path Architecture; and Dell EMC HPC Storage with Intel Lustre filesystem built on the PowerEdge and PowerVault Modular Disk reference architecture.

In total, OzSTAR will boast 4,140 Intel Xeon Scalable Processor cores at 2.3Ghz across 107 standard compute and 8 data crunching nodes; 230 Nvidia Tesla P100 12GB GPUs -- one per CPU socket; 272 Intel Xeon Phi cores at 1.6Ghz across 4 C6320p KNL nodes; and a high-speed, low latency network fabric to allow for moving data at over 100Gbps.

Underwood said each compute node is a building block to provide the processing, data analysis, and visualisation of the data. He said typical supercomputing environments and centres require multiple supercomputers to achieve the compute, data analysis, and visualisation power that OzSTAR holds.

"Data movement with astronomy is one of the most expensive things; when you're trying to take petabytes of data and move it across the country, there's a cost always involved with that, not just the financial cost, but the processing time," Underwood explained.

The supercomputer is expected to deliver more than 1 petaflops of performance capability -- which equates to 31 million years of calculations in a single second.

It will also have 5 petabytes of usable storage, and a single 60GB file will be able to be moved from one place to another in one second.

OzSTAR is expected to take four weeks to install and will be operational before the end of September.

Once up and running, up to 35 percent of the supercomputer's time will be spent on OzGrav research related to gravitational waves. The supercomputer will also continue to incorporate the GPU Supercomputer for Theoretical Astrophysics Research (gSTAR), operating as a national facility for the astronomy community funded under the federal National Collaborative Research Infrastructure Scheme in cooperation with Astronomy Australia Limited.

In addition, the supercomputer will underpin the research goals of Swinburne staff and students across multiple disciplines, including molecular dynamics, nanophotonics, advanced chemistry, and atomic optics.

Last year, EMC built a private OpenStack research cloud for Swinburne University that Underwood said will be directly connected via Ethernet into the OzSTAR supercomputer to make sure the data movement is optimised.

The cloud forms part of the National eResearch Collaboration Tools and Resources project (NeCTAR).

The CSIRO welcomed a new supercomputer to its Canberra campus last month, with Dell EMC providing the new Bracewell system that is expected to expand CSIRO's capability in deep learning, further its artificial intelligence progress, and allow for the exploration of virtual screening for therapeutic treatments, traffic and logistics optimisation, modelling of new material structures and compositions, machine learning for image recognition, and pattern analysis.

Dell EMC was also awarded a AU$1.2 million contract for the expansion of CSIRO's Pearcey supercomputing system earlier in July.

Read more from the original source:

Swinburne reaches for the sky with AU$4m astronomical supercomputer - ZDNet

The computers the human race currently work with aren't built for space, and can't last long in off-planet environments. Astronauts aboard the ISS need new machines every month, as the old ones get destroyed by various factors that don't exist on Earth, like solar flares, radiation, subatomic particles, and irregular cooling.

Considering that SpaceX is currently preparing for a mission to Mars, it is paramount that the astronauts that eventually head out there have durable, working machines. Now,HP has announcedit is working with NASA to create such a computer.

The HPE Apollo Spaceborne Computer is going to be lifted into space, on board the SpaceX-12 rocket. It's a "high performance commercial off-the-shelf (COTS) computer system." It was ruggedized to withstand the harsh environments of space but not from the hardware side --which is the usual approach --but from the software side instead.

It passed 146 safety tests and certifications before NASA cleared it for space.

"The Spaceborne Computer includes the HPE Apollo 40 class systems with a high speed HPC interconnect running an open-source Linux operating system. Though there are no hardware modifications to these components, we created a unique water-cooled enclosure for the hardware and developed purpose-built system software to address the environmental constraints and reliability requirements of supercomputing in space," HP said.

"We see the Spaceborne Computer experiment as a fitting extension to our HPE Apollo portfolio, purpose-built for supercomputing. HPE is excited to expand its relationship with NASA, pioneering HPC in space and taking one step closer to the Mission to Mars."

Published under license from ITProPortal.com, a Future plc Publication. All rights reserved.

Photo Credit: ESA & MPS for OSIRIS Team MPS/UPD/LAM/IAA/RSSD/INTA/UPM/DASP/IDA

Follow this link:

HPE and NASA sending a supercomputer to space - BetaNews

One of the largest challenges a human mission to deep space would face would be the inevitable communications delay caused by the vast distances between Mission Control on Earth, and the spacecraft carrying the crew. During the Apollo missions, in which American astronauts visited the moon, that communications delay was only around 1.3 seconds each way, meaning that youd need to wait 2.6 seconds (1.3 for your message to get there, and 1.3 for their response to come back) to get an answer to your question.

Insignificant as that delay may seem, it will eventually grow to a full 90 minutes or so for the crew of a trip to Mars, meaning the astronauts would not be able to rely on the collective expertise and computing power offered by our ground-based space infrastructure. Emergency course corrections, in three dimensions and with limited fuel, would require the ability to instantly complete complex calculations with little to no margin for error. While there are rumors of early Gemini astronauts doing just that at least once, a mission to Mars or further would require the ability to make these decisions near-instantly; something no human being may be able to do.

or Log In

If you liked this article, tell someone about it

Alex Hollings Alex Hollings served as an active duty Marine for six and a half years before being medically retired from service. As an athlete, Hollings has raced exotic cars, played Marine Corps football and college rugby, fought in cages, and even wrestled alligators. As a scholar, he has earned a masters degree in Communications from Southern New Hampshire University, as well as undergraduate degrees in Corporate and Organizational Communications and Business Management.

Continue reading here:

SpaceX set to deliver 'supercomputer' to the International Space Station for testing - SOFREP (press release) (subscription)

Hewlett Packard Enterprise is sending a supercomputer to the International Space Station aboard SpaceXs next resupply mission for NASA, which is currently set to launch Monday.

Typically, computers used on the ISS have to be hardened, explained Dr. Mark Fernandez, who led the effort on the HPE side as lead payload engineer. This process involves extensive hardware modifications made to the high-performance computing (HPC) device, which incur a lot of additional cost, time and effort. One unfortunate result of the need for this physical ruggedization process is that HPCs used in space are often generations behind those used on Earth, and that means a lot of advanced computing tasks end up being shuttled off the ISS to Earth, with the results then round-tripped back to astronaut scientists in space.

This works for now, because communication is near-instantaneous between low-Earth orbit, where the ISS resides, and Earth itself. But once you get further out as far out as Mars, say communications could take up to 20 minutes between Earth and spaceship staff. If you saw The Martian, you know how significant a delay of that magnitude can be.

3D Model of the supercomputer being sent to the ISS.

3D Model of the supercomputer being sent to the ISS.

3D Model of the supercomputer being sent to the ISS.

3D Model of the supercomputer being sent to the ISS.

Suppose theres some critical computations that need to be made, on the mission to Mars or when we arrive at Mars, we really need to plan ahead and make sure that that computational capacity is available to them, and its not three- or five-year old technology, Dr. Fernandez told me. We want to be able to get to them the latest and greatest technology.

This one-year experiment will help Dr. Fernandez hopefully show that in place of hardware modifications, a supercomputer can be software hardened to withstand the rigors of space, including temperature fluctuations and exposure to radiation. That software hardening involves making adjustments on the fly to things like processor speed and memory refresh rate in order to correct for detected errors and guarantee correct results.

All of our modern computers have hardware built error correction and detection in them, and its possible that if we give those units enough time, they can correct those errors and we can move on, Dr. Fernandez said.

Already on Earth, the control systems used to benchmark the two experimental supercomputers sent to the ISS have demonstrated that this works a recent lightning storm struck the data center where theyre stationed, causing a power outage and temperature fluctuations, which did not impact the results coming from the HPCs. Dr. Fernandez says thats a promising sign for how their experimental counterpartswill perform in space, but the experiment will still help show how they can react to things you cant test as accurately on Earth, like exposure to space-based radiation.

The SpaceX Falcon 9 rocket, with the Dragon spacecraft onboard, launches from pad 39A at NASAs Kennedy Space Center in Cape Canaveral, Florida, Saturday, June 3, 2017. Dragon is carrying almost 6,000 pounds of science research, crew supplies and hardware to the International Space Station in support of the Expedition 52 and 53 crew members. The unpressurized trunk of the spacecraft also will transport solar panels, tools for Earth-observation and equipment to study neutron stars. This will be the 100th launch, and sixth SpaceX launch, from this pad. Previous launches include 11 Apollo flights, the launch of the unmanned Skylab in 1973, 82 shuttle flights and five SpaceX launches. Photo Credit: (NASA/Bill Ingalls)

And while the long-term goal is to make this technology useful in an eventual mission to Mars, in the near-term it has plenty of potential to make an impact on how research is conducted on the ISS itself.

One of the things thats popular today is Lets move the compute to the data, instead of moving the data to the compute, because were having a data explosion, Dr. Fernandez explained. Well thats occurring elsewhere as well. The ISS is bringing on board more and more experiments. Today those scientists know how to analyze the data on Earth, and they want to send the data down to Earth. Its not a Mars-type latency, but still youve got to get your data, got to process it and got to get back and change your experimental parameters. Suppose, like at every other national lab in the nation, the computer was right down the office from you.

Local supercomputing capability isnt just a convenience feature; time on the ISS is a precious resource, and any thing that makes researchers more efficient has a tremendous impact on the science that can be done during missions. Japans JAXA space agency recently started testing an automated camera drone on board for just this purpose, for instance, since not having to hold a camera means researchers can spend more time on actual experimentation. For HPC, this could have an even larger impact in a couple of different ways.

JAXAs Jem Int-Ball drone camera is also designed to save astronauts time.

You do your post-processing right down the hall on the ISS, youve saved time, your science is better, faster and you can get more work out of the same amount of experimental time, Dr. Fernandez said. Secondly, and more important to me, the network between the ISS and Earth is precious, and its allocated by experiment. If I can get some people to get off the network, that will free up bandwidth for people who really need the network.

In the end, Dr. Fernandez is hoping this experiment opensthe door for future testing of other advanced computing techniques in space, including Memory-Driven computing. He also hopes it opens the door for NASA to consider making use of the same sort of system on future Mars missions, to help with the experimentation potential of those journeys, and to help improve their chance of success. But in general, Dr. Fernandez says, hes just hoping to contribute to the work done by those advancing various fields of research in space.

I want to help all the scientists on board, and thats one of the dreams of this experiment, he said.

Read this article:

Why HPE is sending a supercomputer to the ISS on SpaceX's next rocket - TechCrunch

Tomorrow, SpaceXs Falcon 9 rocket is set to launch another batch of cargo and science experiments to the International Space Station, and that shipment will include a supercomputer from Hewlett Packard Enterprise. Called the Spaceborne Computer, the system is a joint project between HPE and NASA to see if a commercial computer can be designed to last in the harsh space environment. If successful, similar computers could be critical tools for future deep-space missions beyond Earth.

The space stations location in lower Earth orbit makes it an unfriendly place for computers. Because the ISS sits outside the majority of Earths protective atmosphere, its exposed to more radiation from solar flares and cosmic rays that originate outside the Solar System. This exposure can degrade technology over time, so computers that go to space have to be physically hardened with shielding in order to withstand this higher radiation environment. But this upgrading process takes a lot of time and money, and it adds weight to the computer, according to HPE.

Can a commercial computer last in the harsh space environment?

The Spaceborne Computer is an experiment to see if regular off-the-shelf computers can operate in space over long periods of time. The computer is also equipped to deal with radiation exposure differently, relying on software upgrades rather than hardware. It runs on an open-source Linux operating system and is programmed to recognize when a high-radiation event is occurring, for instance. It will then respond by throttling its systems and lowering its operating speed to save power and avoid damage, according to NASA.

The computer supposedly passed 146 safety tests and certifications to be approved for space travel by NASA, says HPE. Once it launches to orbit, the computer is supposed to last a year on the ISS. Overall, NASA wants to know just how much the computer will suffer from radiation exposure over time, and if these software patches can actually reduce any degradation. The results of the space-bound computer will be compared to an identical computer that HPE is keeping on the ground.

If the experiment works, similar software-hardened computers could be critical for future missions to Mars. Communicating with astronauts on the Red Planet will be a slow process, since a radio signal takes around 20 minutes to travel from Earth to Mars. That means round-trip communications could take upwards of 40 minutes. If Mars astronauts have to do complicated calculations in a hurry, they wont be able to rely on Earth; theyll have to use any computers they take with them and those systems will need to be able to withstand the heightened radiation on the way to Mars.

SpaceXs Falcon 9 rocket is slated to take off at 12:31PM ET from Cape Canaveral, Florida, with the Spaceborne Computer onboard.

Here is the original post:

SpaceX's next launch will send an HPE supercomputer to the International Space Station - The Verge