In our virtual 4 meter below sealevel studio in Almere, we had a guest from the Irish Centre for High-End Computing (ICHEC) in the interview seat to discuss the activities that are happening there and the developments in the context of the European supercomputing ecosystem. Venkatesh Kannan is Technical Manager at ICHEC, Ireland's national supercomputing centre.
To start with, can you tell a little bit about yourself?
Venkatesh Kannan: Absolutely. My name is Venkatesh Kannan. I work as technical manager at the Irish Centre for High-End Computing (ICHEC). I should mention that we are the national centre for high-performance computing in Ireland. The core mission of ICHEC is to deliver HPC-related capabilities and expertise to higher education institutions, enterprises and the public sector on behalf of the state in Ireland.
Can you tell a little bit about the size of the centre? How many people are working there? What is the size of the systems?
Venkatesh Kannan: We are 40+, growing to 45-50 computational scientists at the centre. We are spread across two offices, one in Dublin and the other one in Galway. The National University of Ireland and Galway is our legal host. The centre itself is directly funded by two Irish government departments, one related to employment and innovation and the other one is dedicated to education and skills, and hence the primary candidate to support both academia, public sector and industries in Ireland.
Recently, you had several announcements related to European projects and European Centres. Can you tell a little bit about those?
Venkatesh Kannan: One of the key activities that has recently been very prominent in the European HPC landscape, as you might have seen, is the EuroHPC programme. As a part of that, we recently kick-started the National Competence Centre under EuroHPC: the EuroCC activities. These activities are bringing the expertise in HPC and associated areas to the academic enterprise organisations in the field. These are particularly important because they are the essential driving forces for digital transformation of the scientific industry and public administration organisations in Ireland and also within Europe.
At the moment, we have the EuroCC National Computer Centre involved in two areas. The first one is to support the academic and research communities which is done through the academic flagship programme that has already been kick-started. The objective of the academic flagship programme is to increase the competitiveness of Irish academic groups when they are applying for access to larger European systems. This is done by providing hands-on performance engineering assistance for the researchers to scale up and optimize their application codes. By the end of the two-year academic lecture programme, we expect these research groups and their codes to be ready so that they can be successfully run in an optimized way on European tier-0 PRACE systems or the pre-exascale and exacale systems that are coming up. The scientific domains of these projects span different domains: from astronomy to artificial intelligence for nanoscience, optimization of machine learning and deep learning, and material sciences.
Is the industry also involved?
Venkatesh Kannan: Yes, absolutely. The academic flagship is one aspect of the National Competence Centre. The other key aspect or pillar within the Competence Centre is the SME accelerator. This kind of ties into our ongoing activities in bringing HPC and capabilities to enterprises. The ICHEC SME accelerator particularly works at the convergence of Big Data artificial intelligence and HPC technologies with Irish SMEs and start-ups and public sector organisations in order to accelerate their technology adoption for digitally transforming their services and solutions. The model of engagement is primarily directed towards two aspects. One is raising the technical awareness of the enterprises and public sector organisations as to what HPC and related technologies can bring to them. The second aspect is to accelerate and ease their adoption and use these solutions to build their applications.
One of the projects that you are involved in is the EuroCC project that kicked off very recently. Do you have a special role in that or are you just one of the Competence Centres?
Venkatesh Kannan: We are one of the participants in the EuroCC project. We are one of the 33 national Computing Centres.
Are there other European projects that you are involved in?
Venkatesh Kannan: In addition to the National Competence Centre, I might actually bring your attention to a project that is focusing on the next generation of data management and storage platforms which are going to be key to sit alongside the exascale systems that are going to come up. This project proposal was recently successful. The project is due to kickstart in quarter one of 2021. It is called IO-Sea. It is a three-year project led by CEA in France. The objective of the IO-Sea project is to provide novel data management and storage platforms for existing computing platforms targeting zettabyte- scale storage systems. These are primarily one thousand times exabyte systems. The focus is to develop critical aspects of intelligent placement of extreme volumes of data across different categories and types of storage and memory components that go into the exascale systems. It is particularly important to place the right resources and right fragments of data on different parts of the memory and storage systems. In this way, the computation is optimized both from a performance cost and energy perspective on these extreme scale systems.
This is all classical computing, albeit on the exascale level and very innovative. Are you also doing things like quantum or neuromorphic computing?
Venkatesh Kannan: Yes, particularly quantum computing. As you rightly said, quantum computing is still an essential part of high-performance computing in a general sense. The only thing is that it is in a different paradigm based on quantum mechanical systems. In that respect, the focus of ICHEC is pretty much how we address high-performance computing in a classical sense. It is to understand the capabilities of the quantum computing platform and build the expertise and innovation in how we can develop applications on these quantum computing platforms. To this end what ICHEC has been running since 2018 at the national level is the National Quantum Programming Ireland Initiative.
The primary focus of this initiative consists of three objectives. One is understanding how quantum computing platforms are going to be deployed alongside classical HPC systems because at least for the foreseeable future, quantum computing platforms are seen as accelerators that will be coupled to classical HPC systems. It is important to understand how these classical and quantum systems will be integrated at hardware system software and application workflow levels. That is one pillar of the Quantum Programming Ireland Initiative.
The second pillar is the R&D that we collaboratively perform with a number of partners, both in academic research and industry, to understand how software and applications would be developing on these quantum computing platforms. We have in-house expertise in developing algorithms and applications for different demonstrators and use cases on these hybrid quantum and classical HPC systems.
The third aspect is focusing on a national roadmap for developing skills, both with industry and academia, to be able to leverage quantum computing platforms alongside classic HPC systems.
Do you already have systems or simulators in place? How do you run those applications?
Venkatesh Kannan: By the end of the year, we will have a national quantum learning platform which will be a quantum simulation system that will be coupled to the national supercomputer Kay. This will be available for access to both academic researchers as well as industry to build their hybrid application workflows and also for skill development particularly. This platform will be operational by the end of the year. It is also interesting to note the Pilot Quantum Simulator Call from EuroHPC earlier this year. We have a proposal under consideration with a number of European partners. The focus of this proposal, soon to be announced, is to deploy quantum simulator platforms that will be tightly integrated with PRACE tier-0 HPC systems. The objective is to integrate the systems, develop the software for hybrid classical HPC applications and, if successful, ICHEC would be leading with other partners the development of the demonstrators' use cases and also the system software that allows coupling these systems.
What kind of applications are you looking for?
Venkatesh Kannan: We look at applications on both sides, meaning for scientific purposes and also for industry developments as we know quantum computers are far from production level scale. But in terms of scientific applications, we are looking at physics-based simulations, chemistry simulations and on the application side, we are looking at optimization problems and chemistry problems that have an industry relevance.
And then there is the question that we always need to ask in this context: When will the first quantum computer be there with real applications?
Venkatesh Kannan: That is the million dollar question or probably even more. It is a good sign that for example currently within Europe, the capabilities of quantum platforms are around the order of 100 qubits soon towards the end of the year or early next year. For example, companies like Pasqal have these analogue and digital devices which can operate at one hundred or two hundred qubit levels. What we see is an accelerated capability of these quantum computing platforms. The projection to reach a one thousand order qubit system is around 2025. That is reflected in the Horizon Europe Programme as well. What we need to be conscious of though is that to have these practical scale systems we usually need one hundred thousand or even one million orders or millions of qubits. Possibly, it is going to be the next decade when quantum computing might be more practical. Hence, we need to prepare the applications and the algorithms that will be running on these kind of systems.
That is why you see a lot of projects including NEASQC which is the more recently announced quantum flagship project of which ICHEC is also a partner. NEASQC is a four-year project that is coordinated by Atos in France. ICHEC leads the work packages and the tasks on AI and graphical algorithms. The NEASQC project itself focuses on three strands of use cases, 10 in number. The categories of use cases are in chemistry, optimization and in artificial intelligence. ICHEC's focus is on two applications. One is to develop a quantum enabled natural language processing application and a use case with an industry partner. The other application that we are looking at is for chemistry. We are looking at how we can systematically break down larger molecules for their simulation on the smaller scale devices that we have at the moment.
Are there also other architectures that you are looking at? This is a kind of hybrid architecture with classical supercomputing and quantum but are you also looking at other types of architecture?
Venkatesh Kannan: We have in our horizon neuromorphic computing. We recently have been in discussions with various partners in Ireland, such as the Dublin Institute of Advanced Sciences to be able to address those platforms but particularly, we are focusing on classical and quantum computing platforms for the next three-year timeline.
That all sounds very interesting. We are looking forward to the first quantum computing results.
Venkatesh Kannan: Absolutely, it is an exciting time.
Thank you very much for this interview.