Friday, June 29, 2018

Can Russian hackers be stopped? Here's why it might take 20 years



Deterring hackers is almost impossible when the rewards are so great and the risks are so low. Can anything stop them?
By Steve Ranger
In the spring of 2015, faced with external cyberattacks on the US of increasing frequency and severity, President Obama made a dramatic announcement.
The level of hacking and cyber-espionage against the US had created an "unusual and extraordinary threat to the national security, foreign policy, and economy" of the country, said the President, who declared a national emergency to deal with the threat. This executive order allowed the administration to pursue sanctions against those who attacked US critical infrastructure or stole secrets.
Since then the national emergency has been extended three times (it must be reconfirmed every year), but the attacks against the US and its allies continue.
Indeed, the ongoing state of emergency did little to deter the most spectacular anti-US hacking campaign in recent years: Russia's meddling in the 2016 US presidential election.
Russia is not alone in pursuing cyberattacks—which show little sign of stopping—to advance its aims: The US government and its allies have long complained about the behaviour of China, Iran, and North Korea, too.
Despite years of sanctions, indictments, and other attempts to combat hackers, the attacks have continued. And experts have warned it could be 20 years before the situation is brought under control. So why can't the hackers be stopped?

How can governments stop hackers?

It's not that the US hasn't tried to deter cyberattacks, rather that the techniques the country and its allies have used so far haven't been very effective at stopping the bombardment.
"We have seen consistent intrusions and intrusion attempts," said Benjamin Read, manager of the cyber-espionage team at security company FireEye.
Certainly, cybersecurity is a tough concept for politicians to get their heads around. Foreign agents sneaking into computer systems to steal secrets is crazy enough; the idea of enemies hacking into the computers which control critical infrastructure like power stations to cause destruction can seem like something out of an airport thriller—but is scarily real.
But as politicians try to wrap their heads around the concept of "the cyber," in the absence of a clear legal framework, hackers and spy agencies are experimenting to see what they can—and can't—get away with. It's this uncertainty and lack of rules, and any obvious deterrent, that's creating a free-for-all online.
And cyberattacks are cheap, too: No need for a huge military might when all you need is a few smart people and some PCs to start a hacking campaign that can cause headaches for some of the biggest nations on the planet. For a state with few other options, cyberattacks can be a potent weapon.
What makes cyberattacks an even more enticing option is that it's often hard to work out who is actually responsible for a particular incident, making it a handy way to cause trouble without necessarily getting caught. Nations often outsource these kinds of intrusions to freelancers who are adept at covering their tracks, making it harder to point the finger of blame. For example, an intrusion that took French TV station TV5Monde off the air was first thought to be the work of the "Cyber Caliphate" linked to ISIS, but is now blamed on Russia-backed hackers who deliberately left a false trail.
Complicating matters even further is the often-forgotten reality that all countries commit espionage—even against their own allies—to understand capabilities and intentions (there's also an unwritten rule that defence contractors and government agencies are considered to be fair game when it comes to espionage, digital or otherwise).
For Western governments, making a clear distinction between what they consider standard (if unsavoury) elements of statecraft, ones that they indulge in themselves, and those activities that are deemed less acceptable—industrial espionage, election meddling, destructive cyberattacks, and even cyberwarfare, for example—has proven difficult.
Hacking is cheap, easy, deniable, and everybody is doing it. No wonder it's proving so hard to stamp out.

How does cyber-deterrence work?

This is the complicated backdrop against which Western governments are struggling to build some kind of model to deter cyberattackers.
"What deterrence is fundamentally about is making the cost of doing something too high for someone to want to do it," says Ewan Lawson, senior research fellow for military influence at the Royal United Services Institute (RUSI), who was previously with Joint Forces Command with responsibility for the development of cyberwarfarecapabilities.
"The two forms we've tended to see is deterrence through hardening—just making something too difficult to be done—or deterrence through punishment, where you impose a cost on the individual or organisation," he says.
Hardening defences should be the easy part. Many of the most basic attacks—such as the Russian attacks on routers and network infrastructure that the FBI and the UK's GCHQ warned about recently—could be deflected by basic security measures like changing default passwords.
However, while governments have more control over their systems, they have less ability to insist that businesses and individuals improve their own security, which is generally pretty terrible, because there are always better things to do. That means there is always a backdoor open to the hackers—and too often the front door, too.
According to one estimate, more than two-thirds of the UK's critical infrastructure bodies suffered an IT outage in the last two years, a third of which were likely due to cyberattacks. Few companies can survive a sustained assault by hackers, and even fewer are prepared to defend against state-backed attacks.
"The unfortunate reality is that, for at least the coming five to ten years, the offensive cyber capabilities of our most capable potential adversaries are likely to far exceed the United States' ability to defend and adequately strengthen the resilience of its critical infrastructures," warns a 2017 US Department of Defense report on cyber deterrence.
Attempts to tackle the other side of that deterrence equation—imposing a cost on the hackers—have proved even more complicated.
Earlier in 2018 the US director of national intelligence Dan Coates warned that Russia, China, Iran, and North Korea "will work to use cyber operations to achieve strategic objectives unless they face clear repercussions for their cyber operations."
However, finding a set of effective deterrents remains at best a work in progress.
Some state-backed hackers are looking for trade secrets, some are looking for weaknesses that could be used in future attacks, some are looking to steal money—and others want to just stir up trouble. Some want to do all of these things at once. Each of these motivations requires a different response.
"In order to have effective deterrence from a US standpoint it's very important that we not just think about this in terms of cybersecurity defence and offence, but the cultural aspects of various nation states and their motivation," said Trevor Rudolph, a New America cybersecurity fellow who was chief of the Cyber and National Security Division at the Office of Management and Budget during the Obama administration.
Over the last half-decade, the US and its allies have tried to deter state-backed hackers with everything from publicity to sanctions and indictments, and (maybe) even attempts to hack back against assailants.
While governments have plenty of practice at responding to a traditional armed assault because they've been dealing with that pretty much since countries were invented, calibrating a response to a cyberattack remains tricky.
"Ultimately it's not about responding to a cyberattack with cyber means, it's about looking at the full toolkit you have as a state in terms of diplomatic, economic, military, and others, and determining the right set of incentives and penalties you're going to apply to a country that's behaving in a way that is unacceptable," says Dmitri Alperovitch, CTO at security company CrowdStrike

Cyber-deterrence trial and error

The US, in particular, has been testing a variety of different deterrent strategies over a number of years. China was the first country openly tackled for its cyber-espionage when, in May 2014, a grand jury indicted five Chinese military hackers for hacking directed at companies in the US nuclear power and solar energy industries (see timeline below).
While attempts to curb the behaviour of China, Iran, and North Korea has been limited in its impact, the biggest challenge the US faces at the moment is from Russian interference.
A summit between President Obama and Chinese President Xi Jinping followed a year later, at which both countries promised not to use commercial cyber-espionage. Chinese attacks slowed, at least temporarily. But, according to the US intelligence community, China continues to use cyber-espionage to try and break into defence contractors and communications firms in particular.
China is also targeting confidential business information such as pricing strategies or mergers and acquisitions data says FireEye's Read. "What we've seen pop up is Chinese groups targeting US law firms, US investment companies, and so on, stealing information in support of economic goals."
Attempts to curb cyber intrusions by Iran have also met with similar, limited, success.
In March 2016 charges were announced against seven Iranians over distributed denial of service attacks against US companies; one man was also charged with unauthorized access into control systems of a US dam. In March 2018 the US Department of Justice charged nine Iranians with stealing more that 31 terabytes of documents and data from more than 140 American universities and 30 American companies.
The US also tried using sanctions against North Korea over its hacking attack on Sony Pictures in what was the first use of sanctions by the US in response to cyber-espionage.
It's possible, although still unclear, that the US may have also responded to Pyonyang's attack on Sony Pictures by taking North Korea's internet offline for a short period of time, but even this has done little to curb North Korea's activities.
But North Korea continues to use cyberattacks to gain intelligence and in particular to steal funds to prop up the state. "They've really veered into the crime angle," Read notes.
While attempts to curb the behaviour of China, Iran, and North Korea has been limited in its impact, the biggest challenge the US faces at the moment is from Russian interference.
Russia has been blamed for the hacking of the Democratic National Committee and the subsequent leaking of emails. Kremlin-backed groups have also been accused of using disinformation campaigns across social media to stage arguments and undermine trust in the US political system during the 2016 Presidential campaign.
For its part, Russia has denied any meddling. President Putin has denied Russian state involvement in any election meddling, although he did not rule out that Russian hackers might be involved.
"If they are feeling patriotic, they will start contributing, as they believe, to the justified fight against those speaking ill of Russia," he told journalists in 2017. But then, in March 2018 Putin again denied Russian state involvement: "Why have you decided the Russian authorities, myself included, gave anybody permission to do this?" he told NBC News.
US intelligence warns that Russian intelligence and security services continue to probe US critical infrastructures, as well as target the US, NATO, and allies for insights into US policy. Attempts to deter Russian meddling seem to have had little impact.
"It has not changed the calculus or the behaviour on behalf of the Russians," said then-NSA chief Admiral Mike Rogers in February 2018 according to CNN. "They have not paid a price that is sufficient to change their behaviour," he added.
In December 2016 President Obama responded to revelations about Russian behaviour by expelling diplomats and closing two Russian properties. President Trump added to those moves with new sanctions in March 2018—which had been approved by Congress seven months earlier—and accused Moscow of attempting to hack the US energy grid. Critics said these sanctions did not go far enough.
"The sanctions... are a grievous disappointment and fall far short of what is needed to respond to that attack on our democracy," said Adam Schiff, senior Democrat on the House of Representatives Intelligence Committee (more sanctions were imposed on Russia in May 2018).
Indeed, deterring Russia is further complicated by Donald Trump's own response to the hacking revelations. In the Presidential race he—jokingly—invited Russia to hack Hillary Clinton, saying: "Russia, if you're listening, I hope you're able to find the 30,000 emails that are missing." And after winning the election he was initially reluctant to blame Russia for election meddling.

The limits of naming-and-shaming

One tactic the US has used with some success is to be more public about Russian attacks; it has also coordinated with other countries to go public.
"The US government using its own sources and methods coming out solo in saying 'This nation state did a bad thing' will have a debatable effect. What I think is more effective is if an international body or a group of nations can come out and conclusively say that a particular nation state or group did something," said Rudolph.
"If they are feeling patriotic, they will start contributing, as they believe, to the justified fight against those speaking ill of Russia."
VLADIMIR PUTIN
In February 2018, seven nations—the US, the UK, Denmark, Lithuania, Estonia, Canada, and Australia—blamed the NotPetya ransomware attacks on Russia, with support from New Zealand, Norway, Latvia, Sweden, and Finland. Similarly, it was the US along with the UK and Australia in April 2018 that complained about Russian interference with routers and internet infrastructure.
Creating a broader coalition makes its condemnation stronger and harder for a country to shrug off.
But although naming-and-shaming may have worked against Chinese industrial cyber-espionage (at least in the short term), it doesn't seem to be particularly effective against the Russians. While the Chinese government doesn't like to be embarrassed in this way, Russia seems much less concerned.
While Moscow consistently denies conducting any of these attacks, it doesn't seem to mind the accusations too much—if only because it acknowledges the Russian state's capabilities.
"If you look at the Russian government, you have to pair unilateral sanctions with international sanctions and the real threat of retribution. What we did in the last [Obama] administration in response to the election meddling was insufficient at best," said Rudolph.
Some analysts go further: "Building cyber-deterrence through a mix of both national capabilities and global norms that guide behavior has been a cornerstone of US cyber security since the very realm first emerged. Today, it is not just challenged, but in utter collapse," said cybersecurity expert Peter Singer in a January 2018 article.

So what would a stronger response look like? RUSI's Lawson suggests that targeting Russian oligarchs and their wealth could be effective. "All of this comes with a degree of risk, and one of the things we don't do very well these days is accept risk. None of these things have a 100% guaranteed outcome, but the do-nothing option is not an option either," he says.
"At the moment for Russia there are no consequences. For Putin it all seems to be working," said Lawson.
So has there been any deterrent effect of Western governments' measures so far against Russia? "Not one that I've heard communicated in a very effective way, and certainly not one when I've spoken to Russians that I think they would recognise as such," says Lawson.
But Rudolph sees some positive developments. The US plans to invest more in identifying who is really behind an attack. The National Security Strategy published in 2017 specifically mentioned plans to invest in capabilities that improve the ability of the US to attribute cyberattacks more effectively, and there is now more of a recognition that the one-size-fit-all approach to cyber-deterrence doesn't work.
"That's something I've seen the Trump administration start to evolve, and I do believe they have different response packages, and for different nation states and that's not something we always had," Rudolph said.

Where does cyber-deterrence go next?

There is always the chance that nation states will change their minds about their use of hacking and cyber intrusion.
As recently as 2009 Russia was keen for a treaty with the US covering the use of cyberweapons. This would have banned countries from embedding code in the systems of other nations and imposed a ban on the use of deception to disguise the source of cyberattacks. The US wasn't interested, however. President Trump has also floated the idea in 2017 that the US and Russia create an "an impenetrable Cyber Security unit" to prevent election hacking, but this didn't get very far.
"I'm sceptical as to whether we can develop a meaningful or effective deterrent strategy," said Rudolph.
It will likely take years, or even decades, for rules to finally emerge that govern cyber-espionage and cyberwarfare, so countries will continue to jockey for position for years to come until norms are established. A failure to establish boundaries accepted by all means that the risk of accidental escalation remains; if the rules of engagement aren't clear, then a relatively trivial hacking incident could rapidly turn into a full-on confrontation.
More worryingly, the kind of cyberattacks that governments are required to deter may also change over time. As societies become more reliant on technology—like the Internet of Things—the risk of catastrophic cyberattacks will rise.
"A large-scale cyberattack on civilian critical infrastructure could cause chaos by disrupting the flow of electricity, money, communications, fuel, and water. Thus far, we have only seen the virtual tip of the cyberattack iceberg," said a 2017 report on cyber deterrence.
Governments might choose to deter the largest attacks with the largest weapons in their military arsenals. It was recently suggested that a sufficiently serious cyberattack could result in a nuclear response, which not everyone thinks is far-fetched.
"In order to have effective deterrence from a US standpoint it's very important that we not just think about this in terms of cybersecurity defence and offence, but the cultural aspects of various nation states and their motivation."
TREVOR RUDOLPH
"I think the national security community has lacked a bit of creativity when it comes to thinking of the sheer scope and scale of a massive cyberattack, and the physical consequences that come from that, so I think the nuclear deterrent is interesting—it's part of the evolution of what is going to be a truly effective deterrence. We just don't know yet, and I don't think we will know for about 20 more years," said Rudolph.
One further complication is that rival countries have very different definitions of national security and how to protect it—understanding these differences will be key to creating an agreed set of rules. This makes cyberwar a question of language, not computer code.
RUSI's Lawson argues the West's adversaries aren't playing by the same rules "so surely it makes sense to continue the conversation and at least start to explore where the boundaries lie."
For example, Russia is—among other things—very concerned about the ability of the West to influence its population through the internet in the way that it did in the past through radio stations, and sees its own election meddling as acceptable through that prism of suspicion.
"It's about continuing the conversation," said Lawson. "If it does take 20 years for norms to appear in part that will be our fault for making the decision not to engage."
But for now, many nations states will judge that using hackers to spy on, disrupt, distract, and steal from rival states remains a cheap, effective, and relatively risk-free option. Until something changes, expect to see plenty more of the same.

A brief history of cyber-deterrence

In the last few years the US government and others have made various attempts to stop hacking attacks by countries including Russia, China, North Korea, and Iran, with mixed results.
May 2014
A grand jury indicted five Chinese military hackers for computer hacking, economic espionage, and other offences directed at companies in the US nuclear power, metals, and solar energy industries. This was the first time charges were levelled at state-sponsored hackers for economic espionage. "For too long, the Chinese government has blatantly sought to use cyber espionage to obtain economic advantage for its state-owned industries," said then-FBI Director James Comey.
December 2014
Internet service in North Korea is disrupted shortly after President Obama said the US would respond to North Korea's attack on Sony Pictures "in a place and time and manner that we choose." If this was a US attack, it would be the first publicly known retaliation by the US against a cyberattack.
January 2015
US government imposes sanctions on North Korea following its "destructive, coercive cyber-related actions during November and December 2014." This is the first time sanctions are used to respond to a cyberattack.
April 2015
President Obama declares a national emergency to deal with cyberattacks, saying: "The increasing prevalence and severity of malicious cyber-enabled activities originating from, or directed by persons located, in whole or in substantial part, outside the United States constitute an unusual and extraordinary threat to the national security, foreign policy, and economy of the United States." The executive order authorises a set of new sanctions against individuals or groups whose cyberattacks result in significant threats to the US, and gives authorities the power to freeze assets or apply sanctions against companies that knowingly use stolen trade secrets.
September 2015
Agreement on commercial cyber-espionage. President Obama and China's President Xi Jinping agree that "neither country's government will conduct or knowingly support cyber-enabled theft of intellectual property, including trade secrets or other confidential business information, with the intent of providing competitive advantages to companies or commercial sectors."
March 2016
Charges are announced against seven Iranians for conducting a coordinated campaign of DDoS attacks against 46 companies, mostly in the US financial sector, from late 2011 through mid-2013. One man was also charged with gaining unauthorized access into the Supervisory Control and Data Acquisition (SCADA) systems of the Bowman Dam, in Rye, NY, in August and September of 2013.
December 2016
US places sanctions on Russia over election meddling.
"Russia's cyber activities were intended to influence the election, erode faith in U.S. democratic institutions, sow doubt about the integrity of our electoral process, and undermine confidence in the institutions of the U.S. government. These actions are unacceptable and will not be tolerated," the White House said. The April 2015 executive order is extended to authorise sanctions against those who: "Tamper with, alter, or cause a misappropriation of information with the purpose or effect of interfering with or undermining election processes or institutions."
February 2017
The Department of Defense's Defense Science Board Task Force on Cyber Deterrence warns: "It is clear that a more proactive and systematic approach to U.S. cyber-deterrence is urgently needed."
March 2017
The national emergency is extended. "The President believes that the significant cyber-enabled activities continue to pose an unusual and extraordinary threat to our national security and economic prosperity, and therefore he has determined that it was necessary to continue this national emergency," said then-White House press secretary Sean Spicer.
November 2017
Three Chinese nationals are indicted for computer hacking, theft of trade secrets, conspiracy, and identity theft directed at US and foreign employees and computers of three corporate victims in the financial, engineering, and technology industries between 2011 and May 2017.
December 2017
The US National Security Strategy says the country will "impose swift and costly consequences on foreign governments, criminals, and other actors who undertake significant malicious cyber activities." It adds: "We will also invest in capabilities that improve the ability of the United States to attribute cyberattacks."
February 2018
The US and the UK blame Russia for NotPetya ransomware, saying: "It was part of the Kremlin's ongoing effort to destabilize Ukraine and demonstrates ever more clearly Russia's involvement in the ongoing conflict. This was also a reckless and indiscriminate cyberattack that will be met with international consequences."
March 2018
More US sanctions on Russia following election meddling and NotPetya. "The Administration is confronting and countering malign Russian cyber activity, including their attempted interference in US elections, destructive cyberattacks, and intrusions targeting critical infrastructure," said Treasury Secretary Steven Mnuchin.
March 2018 
The US Department of Justice charges nine Iranians with conducting a massive cyber theft campaign, stealing more that 31 terabytes of documents and data from more than 140 American universities and 30 American companies.
March 2018
Cyber state of emergency extended again. "Significant malicious cyber-enabled activities originating from or directed by persons located, in whole or in substantial part, outside the United States continue to pose an unusual and extraordinary threat to the national security, foreign policy, and economy of the United States," says President Trump.
April 2018
The US, the UK, and more accuse Russia of hacking into networks. "We have high confidence that Russia has carried out a coordinated campaign to gain access to enterprise, small office, home office routers known as SOHO routers and residential routers, and the switches and connectors worldwide," said Rob Joyce, White House cybersecurity coordinator.
May 2018
The US State Department publishes a document on cyber-deterrence that warns "Strategies for deterring malicious cyber activities require a fundamental rethinking."
June 2018
The US Treasury Department announces further sanctions against five Russian companies and three individuals, part of its attempt to tackle "Russia's malign and destabilizing cyber activities." The Treasury said the sanctions targeted Russia's cyber and underwater capabilities, and said "Russia has been active in tracking undersea communication cables, which carry the bulk of the world's telecommunications data."

Thursday, June 28, 2018

The world's 25 fastest supercomputers


US back on top
The latest TOP500 list of the fastest supercomputers in the world is out, and while the US is back in pole position for the first time in six years, Chinese machines continue to dominate the wider list.
The new TOP500 list is led by Summit, a US Department of Energy supercomputer capable of performing 122 quadrillion floating-point operations per second.
While there are many changes in the top 25, most of machines still rely on vast numbers of Intel CPUs, aided by Nvidia GPUs or many-core Intel Xeon Phi processors.
Here are the 25 most powerful machines in the TOP500 list of supercomputers.

Here we go:


1. Summit


Summit is housed at Oak Ridge National Laboratory (ORNL), where it aids in research into subjects like material science, cancer, and fusion energy for the US Department of Energy (DoE).
The last time a US supercomputer topped the TOP500 list of the world's fastest supercomputers was in 2012, when the spot was held by the Titan supercomputer, also based at ORNL.
The machine has 4,356 nodes, each equipped with two 22-core Power9 CPUs, and six Nvidia Tesla V100 GPUs. Its nodes are linked using a Mellanox dual-rail EDR InfiniBand network.
Processor cores: 2,282,544
Max performance (Linpack benchmark): 122.3PFlop/s
Memory: 2.8PB
Power consumption: 8,805.50kW
Image: Oak Ridge National Laboratory

2. Sunway TaihuLight


The former number one machine, the Sunway TaihuLight until recently set a new standard for supercomputing speed.
Based in the National Supercomputing Center in the city of Wuxi, the Chinese system performs calculations to aid research and engineering work, ranging from climate modelling to advanced manufacturing.
Unlike most other supercomputers, the TaihuLight doesn't rely on Intel CPUs but instead utilises a custom ShenWei processor, a RISC CPU with 260 cores, and custom interconnects made in Wuxi.
Processor cores: 10,649,600
Max performance (Linpack benchmark): 93PFLOPS - quadrillion floating point operations per second
Memory: 1.3PB
Power consumption: 15,371kW
Image: National Supercomputing Center

3. Sierra


Tasked with the crucial role of simulating tests of nuclear weapons in the US stockpile, this new machine is based at the Lawrence Livermore National Laboratory in California.
The supercomputer relies on a mix of IBM Power9 CPUs and Nvidia Volta GPUs and is significantly more capable than the lab's existing Sequoia supercomputer, with Sierra able to sustain four-to-six times the performance and five-to-seven times the workload of the older machine.
Processor cores: 1,572,480
Max performance (Linpack benchmark): 71,610TFlop/s
Memory: 1.4PB
Image: Lawrence Livermore National Laboratory

4. Tianhe-2


Once the fastest supercomputer in the world, the Tianhe-2 is capable of more than 33 quadrillion calculations per second.
Otherwise known as the Milky Way 2, the Tianhe-2 supercomputer memory is based in the National Supercomputer Center in Guangzho, China.
The machine is capable of carrying out a massive number of operations in parallel, spreading tasks between its millions of cores. Each of the machine's nodes has two Intel Xeon E5 Ivy Bridge processors and custom-built Matrix-2000 coprocessors.
Processor cores: 4,981,760
Max performance (Linpack benchmark): 61.4PFLOPS - quadrillion floating point operations per second
Memory: 2.3PB
Power consumption: 18,482kW
Image: Zhao zilong - Imaginechina

5. ABCI


Based at the National Institute of Advanced Industrial Science and Technology in Japan, ABCI (AI Bridging Cloud Infrastructure) is engaged in research and development of artificial intelligence technologies.
As well as being a processing powerhouse, the system also ranks highly for energy efficiency, claiming eighth place in the Green500 list for its 12.054 gigaflops per watt operating performance.
The Fujitsu-built supercomputer is powered by 20-core Intel Xeon Gold processors, along with Nvidia Tesla V100 GPUs.
Processor cores: 391,680
Max performance (Linpack benchmark): 19,880TFlop/s
Memory: 417,792GB
Power consumption: 1,649.25kW
Image: National Institute of Advanced Industrial Science and Technology

6. Piz Daint


The fastest system in Europe and number six machine in the world is Piz Daint, based at the Swiss National Supercomputing Centre.
A recent upgrade to the Cray XC50 research machine doubled its performance, with Nvidia Tesla P100 GPUs added to its cluster of 2.2GHz Intel Xeon E5-2692 CPUs.
The Piz Daint has come a long way, entering the TOP500 supercomputer list at number 114 in 2012, but steadily climbing thanks to repeated upgrades.
Processor cores: 361,760
Max performance (Linpack benchmark): 19.6PFLOPS - quadrillion floating point operations per second
Memory: 340,480GB
Power consumption: 2,272kW
Image: Swiss National Supercomputing Center

7. Titan


Once the world's fastest, this US-based supercomputer slips down to seventh place.
Capable of handling more than 17 quadrillion calculations per second, Titan helps researchers at the US Oak Ridge National Laboratory probe climate change, alternate fuel sources, astrophysics and other major scientific challenges.
The Cray XK7 machine relies on a mix of processors, both AMD Opteron 6274 processors, clocked at 2.2GHz, and NVIDIA K20x accelerators.
Processor cores: 560,640
Max performance (Linpack benchmark): 17.6PFLOPS - quadrillion floating point operations per second
Memory: 710TB
Power consumption: 8,209kW
Image: Oak Ridge National Laboratory

8. Sequoia


Another former frontrunner, Sequoia was not only the most powerful supercomputer in its day but also one of the most efficient.
The IBM Blue Gene/Q system remains a processing powerhouse, and is used used by the US Lawrence Livermore National Laboratory to model massively complex processes, ranging from approximating the universe to the beating of a human heart. It relies on more than one million Power BQC cores.
Processor cores: 1,572,864
Max performance (Linpack benchmark): 17.2PFLOPS - quadrillion floating point operations per second
Memory: 1.6PB
Power consumption: 7,890kW
Image: Lawrence Livermore National Laboratory

9. Trinity


The Trinity supercomputer is another machine engaged in helping the US test the effectiveness of its nuclear arsenal.
Based at Los Alamos National Laboratory, Trinity simulates nuclear explosions using hundreds of thousands of processors.
Processor cores: 979,968
Max performance (Linpack benchmark): 14.1PFLOPS - quadrillion floating point operations per second
Power consumption: 3,844kW
Image: Los Alamos National Laboratory

10. Cori


The flagship machine at the National Energy Research Scientific Computing Center in Berkeley, California, Cori was a new entry in the list last year.
Named after the pioneering biochemist Gerty Cori, the system runs on a mix of Intel Xeon Haswell processors and Intel Xeon Phi many-core CPUs.
Processor cores: 622,336
Max performance (Linpack benchmark): 14PFLOPS - quadrillion floating point operations per second
Memory: 878,592GB
Power consumption: 3,939kW
Image: National Energy Research Scientific Computing Center

11. Nurion


The research system is a type of Cray CS500 supercomputer, seen above, and relies on a mix of 68-core Intel Xeon Phi co-processors and Intel Omni-Path interconnects.
Processor cores: 570,020
Max performance (Linpack benchmark): 13,929.3TFlop/s
Image: Cray

12. Oakforest-PACS


Formerly Japan's fastest supercomputer, the machine is based at Joint Center for Advanced High Performance Computing in Kashiwa.
Dropping slightly down the rankings this year, the Oakforest-PACS relies on an array of 68-core Intel Xeron Phi 7250 processors.
Processor cores: 556,104
Max performance (Linpack benchmark): 13.6PFLOPS - quadrillion floating point operations per second
Memory: 919,296GB
Power consumption: 2,719kW
Image: University of Tokyo

13. HPC4


Hosted outside of Milan, this Spanish machine will be used for seismic and petroleum-system modelling.
Noted for its energy efficiency, the supercomputer uses a mix of 24-core Intel Platinum 8160 CPUs and Nvidia Tesla P100 GPUs.
Processor cores: 253,600
Max performance (Linpack benchmark): 12,210TFlop/s
Memory: 304,320GB
Image: Eni S.p.A.

14. Tera 1000-2


The Bull Sequana X1000 supercomputer has been installed at the French Alternative Energies and Atomic Energy Commission. Its installation is part of the Tera 1000 project to eventually develop an exascale supercomputer.
Processor cores: 561,408
Max performance (Linpack benchmark): 11,965.5 TFlop/s
Memory: 1.6PB
Power consumption: 1,248.00kW
Image: Philippe Stroppa

15. Stampede2


The Texas Advanced Computing Center's latest supercomputer is the Stampede2. It, like the original Stampede built in 2012, is a Dell machine.
Processor cores: 367,024
Max performance (Linpack benchmark): 10.7 TFlop/s
Memory: 736,512GB
Image: Texas ACC

16. K computer


Japan's K computer has been a fixture in the top 25 for many years.
Situated in Kobe's RIKEN Advanced Institute for Computational Science, the K computer's ample processing power helps tackle pressing global challenges in areas ranging from disaster prevention to medical research.
It relies upon more than 700,000 SPARC64 VIIIfx processors, each rated at 2GHz.
Processor cores: 705,024
Max performance (Linpack benchmark): 10.5PFLOPS - quadrillion floating point operations per second
Memory: 1.4PB
Power consumption: 12,660kW
Image: Riken Advanced Institute for Computational Science

17. Mira


Capable of some 8.6 quadrillion calculations per second, Mira plays a pivotal role in research at the Argonne National Laboratory in Illinois.
The IBM BlueGene supercomputer helps US researchers model everything from the performance of jet engines to the inner workings of the human body.
Driving this machine are a host of 16-core, 1.6GHz Power BQC processors.
Processor cores: 786,432
Max performance (Linpack benchmark): 8.6PFLOPS - quadrillion floating point operations per second
Power consumption: 3,945kW
Image: Argonne National Laboratory

18. Marconi


The Marconi supercomputer used by the Italian research consortium Cineca slips down the charts slightly, despite another recent upgrade.
Processor cores: 312,936
Max performance (Linpack benchmark): 8,413.1 TFlop/s
Memory: 455,168GB
Image: Francesco Pierantoni

19. TSUBAME3.0


Used by the Tokyo Institute of Technology for computational modelling and simulations, work on TSUBAME 3.0 was started in August of last year.
Processor cores: 135,828
Max performance (Linpack benchmark): 8,125 TFlop/s
Memory: 137,984GB
Power consumption: 792.08kW
Image: Nvidia / Tokyo Institute of Technology

20. United Kingdom Meteorological Office


It might not have a name, but the UKMO's supercomputer is the 20th fastest in the world. The Cray XC40 was first launched in 2016 and has cracked the top 25 once again in this list.
Processor cores: 241,920
Max performance (Linpack benchmark): 7,038.93 TFlop/s
Memory: 430,080GB
Image: UK Met Office

21. Theta


A recent specs boost helps Argonne's Theta supercomputer retain its place in the top 25.
Processor cores: 280,320
Max performance (Linpack benchmark): 6,920.9 TFlop/s
Image Argonne National Lab

22. MareNostrum


Another relative newcomer to the top 500 list is the Barcelona Supercomputing Center's MareNostrum.
Processor cores: 153,216
Max performance (Linpack benchmark): 6,470.8 TFlop/s
Memory: 331,776 GB
Power consumption: 1,632.00kW
Image: Barcelona SCC

23. JUWELS Module 1


The first module in the JUWELS supercomputer, the German machine consist of about 2,550 compute nodes, each with two Intel Xeon 24-core Skylake CPUs and 96 GiB of memory. About 1 in 50 nodes also use Nvidia Volta GPUs to boost their parallel processing.
Work on adding a second module, optimized for massively parallel workloads, is scheduled to begin in 2020.
Processor cores: 114,480
Max performance (Linpack benchmark): 6,177.7TFlop/s
Memory: 264,096GB
Power consumption: 1,361kW
Image: Forschungszentrum Juelich

24. Pleiades


Venerable in supercomputer years, NASA's Pleiades has been around since 2011. Its place on the list has risen as high as number 7 but amid fierce competition this year drops to 24.
Processor cores: 241,108
Max performance (Linpack benchmark): 5,951.55 TFlop/s
Memory: 917,344 GB
Power consumption: 4,407kW
Image: NASA

25. Japan Meteorological Agency Cray XC50


This latest Japan Meteorological Agency (JMA) machine can manage 10 times more calculations than its predecessor, as well as handling more data.
Alongside being used to help compile everyday forecasts, the JMA computer will help predict extreme weather, such as typhoons and torrential rain.
Processor cores: 135,792
Max performance (Linpack benchmark): 5,730.5TFlop/s
Power consumption: 1,354kW
Image: Japan Meteorological Agency
The U.S is the leading nation in the SUPERCOMPUTER trend and they are the highest exporter of CHIPS and TECHNOLOGY in the industry, no wonder they also have the highest number of SUPERCOMPUTER in the country. watch out for more here as they evolve. thank you.