[Note: This is an article I originally wrote for TOP500 Blog. It is reproduced with permission here.]

While there is always a lot of buzz about the latest HPC hardware architecture developments or exascale programming methods and tools, everyone agrees that in the end the only thing that counts are the results and societal impact produced by the technology. Results and impacts are coming from the scientific and industrial applications running on HPC systems. The application space is diverse ranging from astrophysics (A) to zymology (Z). So the question arises of how to effectively fund development and optimization of HPC applications to make them suitable for current petascale and future exascale systems.

The answer was provided in the European Union (EU) Horizon 2020 (H2020) e-Infrastructures call, Centres of Excellence for computing applications, which was initiated in September 2014. The work would establish a limited number of Centres of Excellence (CoE) necessary to ensure EU competitiveness in the application of HPC for addressing scientific, industrial or societal challenges. The Centres were conceived to be user-focused, develop a culture of excellence, both scientific and industrial, and place computational science and the harnessing of “big data” at the center of scientific discovery and industrial competitiveness. Centres could be thematic, addressing specific application domains such as medicine, life science or energy; transversal, focused on computational science (e.g., algorithms, analytics, and numerical methods); challenge-driven, addressing societal or industrial challenges (e.g., aging, climate change, and clean transport); or a combination of these approaches.

Eight Centres of Excellence for computing applications were subsequently selected for funding and established before the end of 2015. They cover important areas like renewable energy, materials modeling and design, molecular and atomic modeling, climate change, global system science, and bio-molecular research, as well as tools to improve HPC applications performance. Now, nine months later, these Centres are up and running and it is worth to have a closer look at the different ones:

  • CoeGSS – CoE for Global Systems Science will address the emerging scientific domain of Global Systems Science (GSS), which is a vital challenge for modern societies to understand global systems and related policies. The field will use high performance computing as a critical tool to help overcome extremely complex societal and scientific obstacles. Due to the nature of the problems addressed in typical GSS applications, the relevant data sets are usually very large, highly heterogeneous in nature, and are expected to grow tremendously over time.  Bridging HPC with high performance data analysis is thus the key to the success of GSS in the next decade.
  • EoCoE – Energy Oriented CoE is helping the EU transition to a reliable and low-carbon energy supply using HPC. The Centre is focusing on applications in (a) meteorology as a means to predict variability of solar and wind energy production; (b) materials employed for photovoltaic cells, batteries and super capacitors for energy storage; (c) water as a vector for thermal or kinetic energies, focusing on geothermal and hydropower; and (d) fusion for electricity plants as a long-term alternative energy source. These four areas will be anchored within a strong transversal multidisciplinary basis providing expertise in advanced mathematics, linear algebra, algorithms, and HPC tools.
  • E-CAM – Supporting HPC Simulation in Industry and Academia is an e-infrastructure for software, training and consultancy in simulation and modeling. It will identify the needs of its 12 industrial partners and build appropriate consultancy services. E-CAM plans to create over 150 new, robust software modules, directed at industrial and academic users, in the areas of electronic structure calculations, classical molecular dynamics, quantum dynamics, and mesoscale and multi-scale modeling.
  • MaX – Materials design at the eXascale CoE  is supporting developers and end users in materials simulations, design and discovery. It is enabling the best use of HPC technologies by creating an ecosystem of codes, data workflows, analysis, and services in material science to sustain this effort. At the same time, it will enable the exascale transition in the materials domain by developing advanced programming models, novel algorithms, domain-specific libraries, in-memory data management, software/hardware co-design and technology-transfer actions.
  • NOMAD – The Novel Materials Discovery Laboratory is developing a materials encyclopedia and big data analytics toolset for materials science and engineering. The Centre will integrate the leading codes and make their results comparable by converting (and compressing) existing inputs and outputs into a common format, thus making this valuable data accessible (as the NOMAD Repository) to academia and industry. It currently contains over three million entries.
  • BioExcel – CoE for Biomolecular Research is operating towards advancement and support of the HPC software ecosystem in the life sciences domain. Research and expertise covers structural and functional studies of the main building blocks of living organisms (proteins, DNA, membranes, etc.) and techniques for modeling their interactions, ranging from quantum to coarse-grained models, up to the level of a single cell. The Centre will improve the performance, efficiency and scalability of key codes in biomolecular science, make ICT technologies and workflows easier to use, promote best practices, and train end users.
  • POP — Performance Optimisation and Productivity CoE gathers leading experts in performance tools/analysis and programming models in Europe. It is the only transversal CoE. The Centre offers services to the academic and industrial communities to help them better understand the behavior of their applications, suggests the most productive directions for optimizing the performance of the codes, and helps implementing those transformations in the most productive way. The consortium includes academic and supercomputing centers with a long track record of world-class research, as well as service companies and associations with leading expertise in high performance support services and promotion.

Teams from the Jülich Supercomputing Centre are involved in four of the CoE: EoCoE, E-CAM, MaX, and POP (where my team is participating).

Me presenting at the Multicore@Siemens 2016 conference -- Picture by Siemens

Me presenting at the Multicore@Siemens 2016 conference — Picture by Siemens

Yesterday, I gave an invited talk at the Multicore@Siemens 2016 conference in Nürnberg about the performance analysis of parallel applications. While in our high-performance computing center at JSC, we have to deal with large-scale scientific applications running on our world-class very scalable HPC systems like JURECA or JUQUEEN, software developers in general deal with much smaller systems.

Multicore@Siemens 2016 conference — Picture by Bernd Mohr

However, everyone has to deal with parallel (multicore) systems now: smartphones, tablets, or laptops nowadays typically have two or four compute cores and a graphics accelerator and the same is true for embedded computers in consumer devices like washing machines or process automation control systems. Multicore computers are everywhere and so every software developer has to learn and understand parallel programming in these days and quickly finds out: (a) it is complicated to get right and (b) it is even more complicated to make it efficient, that means that the software really makes use of all the computer power available by all the cores on the chip.

If you are interested in the topic, here are the slides. I gave a very similar talk at the Multicore Day 2016 organized by SICS in Stockholm, Sweden and they even videotaped it.

In my talk, I presented some of the results of the RAPID ((Runtime Analysis of Parallel applications for Industrial software Development) project, which is a collaboration between the Corporate Technology Multicore Expert Center of Siemens AG and Jülich Supercomputing Centre.

RAPID (Runtime Analysis of Parallel applications for Industrial software Development) project Logo

RAPID (Runtime Analysis of Parallel applications for Industrial software Development) project logo

The goal of this project was to adapt the measurement and analysis tools Score-P and Scalasca, which we develop at Jülich in my team for many years now, to the needs of industrial applications. As industrial applications are parallelized differently than scientific application codes, it meant that we had to integrate support for threading models like POSIX threads, Windows threads, Qt threads, and ACE threads into Score-P. In addition, support for leveraging task parallelism using MTAPI, the Multicore Association Tasking API,  was also developed. Besides supporting new programming paradigms, additional work had to be done with regards to portability. Although Score-P is already quite portable as it is running on all relevant supercomputer architectures, systems like Windows and operating systems for embedded systems had not been targeted so far, but are of course very important in an industrial context. On the analysis side, new methods targeting thread-based communication patterns, e.g., a lock contention analysis, were implemented in Scalasca. Meanwhile, our software was successfully used in the work of the Multicore Expert Center to understand and optimize important Siemens industry codes.

My new coffee mug from the Siemens Multicore Expert Center -- Picture By Bernd Mohr

My new coffee mug from the Siemens Multicore Expert Center — Picture By Bernd Mohr

At the end of my talk, they gave me a Siemens Multicore Expert Center coffee mug — not sure whether they read my blog article about my coffee mug collection, but anyhow, the mug will get a prominent spot in my bookshelf 😉

Travel for my job often means visiting colleagues all over the world at their high-performance computing centers. As they are hosting some of the most powerful, and therefore very expensive, computers in the world, it is clear that this requires a little bit more than a large enough room in the basement of your university or research center institute. For one thing, they require quite some power: a typical top25 HPC system installation with the actual computer, the high-speed network, enough storage and cooling needs anything between 5 and 25 MW.

In this first part of a new mini-series (“HPC Centres Around the World”), I would like to show you the most beautiful HPC machine room of the world at the Barcelona Supercomputing Center (BSC).

Mare Nostrum Supercomputer at BSC -- picture by BSC (2003)

Mare Nostrum Supercomputer at BSC — picture by BSC (2003)

As you can see, it is actually inside a church, a de-secularized church to be precise 😉  The stories (at least how I remember it) is that in 2003, IBM wanted to demonstrate that it is possible to build a world-class HPC system out if industry-standard compute blades. A partner, the Universitat Politècnica de Catalunya (UPC) at Barcelona, was quickly found. UPC would get a special deal, however, there was the requirement that the system would be up and running in a year. The problem then was to find a building suitable to be able to host a supercomputer, as constructing a new one would have taken too long. The University had a church building on campus which at that time was used for chorus singing. This is actually the reason why the roof of the church inside is covered with a red carpet — it was installed to improve the acoustics for singing. In order to avoid having to cool the whole building a steel / glass frame was constructed inside the church housing the computer. During a sight-seeing tour, visitors are taken to the gallery in the backside of the church, with nice movie theater like seating, with a great view down to the computer. The tour guide can actually walk on top of the glass cage and this way can easily show and explain the different parts of the computer.

If you are interested in more pictures just search for “marenostrum” and “bsc” at Google Images. And, if you visit Barcelona in the future — a good idea anyhow — make sure to reserve some time for the “Temple of HPC Technology” as I call it 😉

BSC machine room from the outside -- Picture by Bernd Mohr

BSC machine room from the outside — Picture by Bernd Mohr

Wow! My own “official” blog! I never thought I would do this one time, but here we go.

[Bernd Mohr]

Me in front of our Jugene supercomputer (2009-2011) – Picture by Ralf-Uwe Limbach

For those who do not know me so well (yet), I am a scientist at the Jülich Supercomputing Centre (JSC) working on Supercomputing, High-Performance Computing and especially performance tools for parallel computing. Besides being researcher, I am also deputy head of the JSC division “Application support”.

I plan (for now) to blog about my research activities and projects, my visits to workshops, conferences and colleagues all over the world, and about my quest to organize SC17.

I am working on performance tools for High-Performance Computing (HPC) for almost 30 years  now, at a time the term “HPC” had not even been invented yet. I have been involved in the development of many open-source performance tools among them TAU, Vampir, KOJAK and currently Score-P and Scalasca. Supercomputers, the biggest and largest computer systems used to solve the world’s toughest problems, are a fascinating research area. The HPC computer hardware architectures, system software and programming models develop so quickly that my work never becomes boring and many exciting research challenges are still ahead of me. If you are interested to learn more about this, and you have some time, you could listen to my podcast about HPC – be aware it is 2.5 hours long and unfortunately, it is in German 🙁

End of 2014, I also got elected the be the General Chair of SC17, the world-largest international conference on high performance computing, networking, storage and analysis attended by over 10,000 people every year. As I am the first non-American after 28 years to organize this conference, it created quite some buzz, for example I made it into the 2015 list “People to Watch” from the online magazine HPCwire. As you can imagine, it is quite an effort to organize a 10,000 attendee multimillion U.S. dollar conference with the help of about 600 volunteers. I will write in the next three years about this effort in a series of blog articles tentatively called “Things you never wanted to learn about SC, but I tell you anyhow!” 😉 If you are interested in this topic, check out the SC15 blog article “10 Questions with SC17 General Chair Bernd Mohr”.

P.S. In case you wonder why the blog is called “Do you know Bernd Mohr?”: The story is that one of our lab directors (name known to the author ;-)) told me once that many times he visits new places or meets new persons, and tells them that he is from Jülich, they often ask him “So you know Bernd Mohr?”.