Top 10 Most Powerful Supercomputers In The World In 2019
Dhir Acharya - Nov 21, 2019
Supercomputers are machines that have a much higher level of performance than a regular computer. And these are the 10 most powerful in the world.
- After Rejection From The US, India Made Its Own Supercomputer, And All Thanks To This Man
- NVIDIA's EGX Supercomputer Is Helping A Bunch Of Tech Giants
- This Superconducting Material Could Power Quantum Computers
By definition, a supercomputer is a computer that has a much higher level of performance than a regular computer, measured in FLOPS (floating-point operations per second) instead of MIPS (million instructions per second). First introduced in the 60s, supercomputers’ performance can now be petaFLOPS (more than a hundred quadrillion FLOPS).
Top500 is a project that ranks and details the world’s 500 most powerful non-distributed computer systems. Starting in 1993, the project updates its list twice per year. And as of now, here are the 10 most powerful supercomputers in the world of this year.
Also known as OLFC-4, this supercomputer was developed by IBM to be used at a lab. Summit runs at a speed of 200 petaFLOPS while its LINPACK benchmark is now clocked at 148.6 petaFLOPS. Additionally, as of last November, it was the world’s third most energy efficient with a power efficiency measured to be 14.668 gigaFLOPS/watt.
It has also become the first supercomputer to reach the speed of exaops, aka a quintillion (10^18) operations per second. It can achieve 1.88 exaops in genomic analysis, expected to achieve 3.3 exaops when using mixed precision calculations.
ATS-2, or Sierra, was built for a lab used by US National Nuclear Security Administration. Primarily, it’s used for predictive applications in the country’s stockpile stewardship, which helps assure the effectiveness, reliability, and safety of the nuclear weapons of the US.
Sierra and Summit have similar architectures, and its speed peaks at 125.712 petaFLOPS, running at a speed of 94.640 petaFLOPS.
3. Sunway TaihuLight
This is a Chinese supercomputer that is rated 93 petaFLOPS on the LINPACK benchmark. Its speed currently peaks at 125.436 petaFLOPS. In June 2017, Sunway TaihuLight was the world’s 16th most power-efficient supercomputer, registering 6.051 GigaFLOPS/watt efficiency.
It used to be the fastest supercomputer in the world between June 2016 and June 2018.
Also called TH-2, this is another Chinese computer making it to this list. It’s currently placed in Guangzhou, China, developed by 1,300 engineers and scientists and has a peak speed of 100.697 petaFLOPS. It currently runs at 61.445 petaFLOPS.
Last year, the US National Science Foundation granted the Texas Advanced Computing Center $60 million to deploy a new computing system called Frontera. The system unlocks new possibilities in the field of engineering and science as it provides the computational capability that enables investigators to cope with more complex, larger research challenges.
Deployed in June this year, the system runs at a speed of 23.516 petaFLOPS, peaking at 38.746 petaFLOPS.
6. Piz Daint
This supercomputer is placed in the Swiss National Supercomputing Centre. It runs at a speed of 21 petaFLOPS and its peak speed is 27.154 petaFLOPS.
Until the end of 2015, Piz Daint was the world’s 8th fastest supercomputer, a higher rank than all other supercomputers in Europe.
Trinity is currently operating in Los Alamos National Laboratory. It ranked 6th on this list in 2015 before dropping to 10th place in November 2016. And it now ranks 7th with a running speed of 20.159 petaFLOPS, peaking at 41.461 petaFLOPS.
8. AI Bridging Cloud Infrastructure
ABCI is a supercomputer built for use in deep learning, machine learning, and artificial intelligence. It runs at a speed of 19.477 petaFLOPS, peaking at 26.874 petaFLOPS.
This machine is placed in the data center of Leibniz Supercomputing Centre, Garching, near Munich. As of summer 2012 when SuperMUC entered operation, it was the fastest supercomputer of Europe.
It serves European researchers across various fields, which include earthquake simulations, genome analysis, life sciences, computational chemistry, computational fluid dynamics, quantum chromodynamics, astrophysics, and medicine.
SuperMUC runs at 19.477 petaFLOPS and peaks at 26.874 petaFLOPS.
This system shares the same architecture with Sierra, but it’s smaller. Its LINPACK benchmark score is 18.2 petaFLOPS while its speed peaks at 23.047 petaFLOPS.