High Performance Computing key to the future

The Ministry of Electronics and Information Technology’s R&D organization Centre for Development of Advanced Computing (C-DAC) and digital transformation company Atos have got together for technology advancement in the areas of quantum computing, Artificial Intelligence and Exascale computing. Here, Arvind Bajaj, Vice President and Head of Big Data and Security Services of Atos India talks about the tie-up.

Can you tell us a bit about your quantum computing tie-up with C-DAC. What benefits can we expect from the quantum simulator and allied research?

The recent agreement between C-DAC and Atos builds on the existing cooperation between the two organizations, whereby a Quantum Learning Machine of 35-Qubits is installed by Atos in C-DAC lab at Pune. The first set of training programs have been conducted, with more advanced sessions planned in the short-term. This is the first step in the long roadmap, which includes multiple milestones culminating in an indigenous quantum computer. This ‘Quantum Computing Experience Center’ that we have jointly setup at Pune, will be an asset to bootstrap the initial research in this field for scientists across the country for specific research in multiple areas like biotech, life sciences, cryptography, etc.

When will the first working quantum computer be ready? How does India currently fare with US and China in quantum research and how can we catch up?

As mentioned, building a quantum computer is a long roadmap, and together with C-DAC, we have reached a milestone with the setup of first quantum computing simulator in India and creation of the lab at C-DAC Pune. At this point in time, there are many unknowns, as the technology is not fully developed and matured. We anticipate this to be a 5-7 year program with the first set of deliverables coming out in the next 3 years or so.

How soon can we develop a supercomputer with processing power in the Exaflop range and how will that benefit the country?

The basic building blocks for Exascale supercomputer exist today, we are looking for ways to optimise the design even further to reduce the energy consumption, physical footprint, among other aspects. Atos is a world leader is Supercomputing and leading the race in Exascale computing. As part of the cooperation agreement, we are working closely with C-DAC to define the architecture and design for its own Exascale system that would leverage the advantages of the next generation technologies expected by next year. Hopefully, we can look at 2021-22 as the launch date for India’s first Exascale system. Such an Exascale/Exaflop system can be imagined to be deployed for the needs of solving social grand challenges around: personalized farming, high-resolution weather forecast, climatology, drug development and personalized medicine, among other areas. Indeed, given the scale of India, in terms of its population, geographical size and diversity, it is conceivable that all the social grand challenges will be requiring an Exascale/Exaflop type of initiatives.

Tell us a bit about your AI research and goals. How will the AI-ML paradigm change when we have an Exascale computer and subsequently a quantum computer?

AI technologies, Machine Learning, Deep Learning and their implementation on large systems like High Performance Computing (HPC) will key for future for many nations. It will lead to, for example, autonomous cars, or new drug discoveries; it will also lead to better offline medicine diagnosis for isolated population. It will also benefit energy or water usage optimization by identifying any outage before it occurs and at the end, save many lives.

The cooperation is broad based and will include multiple areas at infrastructural level (HPC, Quantum Computing Edge Computing) as well as software ecosystem level, with Atos Codex Suite.

Some areas agreed upon with C-DAC, but not limited to, are:

• Using AI to enhance the efficiency of HPC system (Atos project AI4SIM).

• Managing an ecosystem of AI devices such as Edge Computers dedicated to accelerating or concentrating inference execution near the Edge.