Over the years, supercomputers have played a pivotal role in pushing the frontiers of science. Earlier this year, Meta launched one of the fastest AI supercomputers, the AI Research SuperCluster (RSC), to build sophisticated AI models that can learn from trillions of examples; navigate hundreds of different languages; seamlessly analyse text, images, and video together; build AR tools etc.
However, the quest for something even faster than supercomputers led to the development of quantum computers. Last year, the University of Science and Technology of China (USTC) introduced the world’s fastest programmable superconducting quantum computer; Zuchongzhi 2.1 is a million times faster than a conventional computer.
At last year’s I/O conference, Google unveiled a Quantum AI campus in Santa Barbara, California, complete with a quantum data centre, quantum hardware research labs, and quantum processor chip fab facilities. The tech giant plans to build a useful, error-corrected quantum computer within a decade.
Quantum computers of the future will solve complex problems faster and more efficiently than supercomputers. But does it mean supercomputers will become obsolete? Let’s find out.
The first supercomputer came into existence in the 60s. However, the modern supercomputers were developed much later in the 90s. In 1997, Intel developed its first 1 teraFLOPS supercomputer, ‘ASCI red’. Today, the Fugaku supercomputer located at RIKEN Centre for Computational Science in Japan, has thrice the processing power as the world’s second-fastest computer, IBM’s Summit. The Fugaku has clocked a maximum performance of 442,010 teraFLOPs.
Quantum computers, as a concept, were first proposed in the 80s by Richard Feynman and Yuri Manin. In 1998, Isaac Chuang of the Los Alamos National Laboratory, Neil Gershenfeld of MIT, and Mark Kubinec of the University of California built the first quantum computer (2-qubit). In 2017, IBM announced the world’s first quantum computer for commercial use.
“Quantum computing has seen a major boost in the last 10-15 years. Companies worldwide are investing in various quantum technologies and making their quantum hardware.
“Today, we are in the nisq (noisy intermediate-scale quantum) era, working on 100-qubit quantum systems. They may not deliver perfect results (read noisy and erroneous), but you can still work with them. However, we are still very far from achieving the maturity level to have a fully fault-tolerant quantum computer,” said Srinjoy Ganguly, senior data scientist.
Be it IBM’s Sierra or the Sunway TaihuLight, the supercomputers we see today operate at a high compute to I/O ratio. Compared to a conventional computer, a supercomputer runs on multiple processors. The Sunway TaihuLight, one of the top 5 fastest supercomputers globally, has around 40,960 processing modules, each with 260 processor cores.
While a conventional computer works on binary, quantum computers rely on a unit of information called qubits (subatomic particles such as electrons or photons) with far greater processing power. The qubits only work in a controlled quantum state–under sub-zero temperature or in ultra-high-vacuum chambers.
Quantum computing is predicated on two phenomena:
Superposition is the ability of qubits to be in different states simultaneously, allowing them to work on a million computations at the same time. However, qubits are sensitive to their environment, so they can’t maintain their state for long periods. As a result, quantum computers can’t be used to store information long-term.
Einstein described quantum entanglement as spooky action at a distance. It is the ability of two or more quantum systems to become entangled irrespective of how far apart they are. Thanks to the correlation between the entangled qubits, gauging the state of one qubit gives information about the other qubit. This particular property accelerates the processing speed of quantum computers.
Supercomputers are bound by the normal laws of physics. More the processors, the better the speed. Quantum computers are far more efficient than supercomputers as the former harnesses the power of quantum mechanics to carry out calculations. In 2020, China claimed to have developed a quantum computer that performs computations 100 trillion times faster than any supercomputer.
Development and infrastructure cost
Building a supercomputer would cost somewhere between USD 100 million to USD 300 million. For example, the Chinese Sunway TaihuLight cost around USD 273 million. Additionally, the annual maintenance charges fall between USD 4 to 7 million.
Quantum computers are prohibitively expensive. The hardware part alone will cost tens of billions of dollars. The cost per qubit has to come down drastically to make quantum computers commercially viable. At present, a single qubit costs around USD 10,000. Also, qubits operate in a quantum state either in a sub-zero temperature or a vacuum environment, which is very expensive to maintain.
Though a non-quantum algorithm can be run on quantum computers, a quantum algorithm, such as Shor’s algorithm for factoring and Grover’s algorithm for searching an unstructured database, doesn’t work on a supercomputer.
Applications of Supercomputers
Both quantum computing and supercomputing are deployed in cases where large databases are involved. Let’s look at a few use cases:
Weather forecasting: The weather reports we receive on our smart devices come from a supercomputer. Besides predicting the possibility of rain in your city, supercomputers also predict the path of hurricanes and help save thousands of lives.
Last year, the UK’s Met Office signed a 1.2 billion pound deal with Microsoft to develop the world’s most powerful supercomputer to help with preparedness in the face of extreme weather events.
Scientific research: Supercomputers provide insights into complex fields of study. Laboratory experiments are expensive and time-consuming. Hence it is logical to use supercomputers to simulate these laboratory experiments. For example, multiple supercomputers were leveraged across the world to fight the COVID virus and develop vaccines.
Application of quantum computers
“At present, we cannot perform operations on a qubit that lasts more than a few microseconds. Because of this, the quantum data gets lost, making it difficult to be used for AI or other general tasks,” said Ganguly.
Researchers are working on QRAM, a computing unit that will allow storing quantum states for several hours. Quantum computers have applications in fields such as:
Drug design & development: Quantum computers are used to test drug combinations and their interactions. Traditionally, drugs are developed via the trial and error method, which is expensive and risky at the same time.
Computational chemistry: Unlike supercomputers, quantum computers focus on the existence of both 0 and 1 simultaneously, offering immense machine power to map the molecules effectively.
Cryptography: Quantum computers could facilitate secure communications with the help of quantum key distribution. However, there is also a downside.
Recently, US President Joe Biden signed a memorandum asking government agencies to implement quantum-resistant cryptography on their most important systems. The RSA encryption, the most widely used form of encryption, is based on 2048-bit numbers. A quantum computer could break this encryption. As of yet, we don’t have a quantum computer with such capability.