January 09, 2017
CSCS and ETH Zurich strongly encourage the further development of the employees. Among the numerous offers, there is also the sabbatical program. After working for several years, employees in management, staff or support functions may be granted a sabbatical intended to enhance an employee’s professional and personal development and to be of benefit in their future employment at ETH Zurich. The ETH sabbatical can last from a minimum of two months up to a maximum of six months, in agreement with a hosting institution or company, either in Switzerland or abroad, in a relevant field for the professional role of the candidate within the activity at her/his ETH department. Luca Marsella from the Scientific Computing Support group of the User Engagement & Support unit at CSCS had the chance to take his sabbatical at Harvard University. Read his story here:
"In 2016 I participated in the program ETH sabbaticals for Staff, a great opportunity given to ETH employees to enhance their professional and personal development. Since I work within the Scientific Computing Support group of the User Engagement & Support unit at CSCS, the scope of my sabbatical was the exchange of expertise and know-how with the host Institution in the support of high performance applications, with the focus on novel energy efficient architectures for sustainable computing.
I spent my sabbatical from March to May 2016 at the Research Computing of the Faculty of Arts and Sciences (FAS) at Harvard University. Research Computing was established in 2007 to facilitate the advancement of complex research by providing leading-edge computing services. Research Computing staff maintain expertise in constantly changing computing technologies, while speaking the language of FAS researchers, to help them use computing more effectively. Similarly to the corresponding unit at CSCS, Research Computing provides up to date services and technologies to its own academic users, ensuring that researchers have access to a world-class computational environment.
The focus of my visit has been set on the use of scientific applications that employ sustainable power resources and energy efficient large-scale computing infrastructures, thanks to the Massachusetts Green High Performance Computing Center (MGHPCC), a joint venture between Boston University, Harvard University, the Massachusetts Institute of Technology, Northeastern University, and the University of Massachusetts.
Similar to CSCS, MGHPCC is open for use by any research organization and it is managed by the FAS Research Computing in collaboration with the supporting groups of the other founding institutions. MGHPCC provides state of the art infrastructure for computationally intensive research, running several virtual experiments by thousands of researchers in Massachusetts and around the world.
Nowadays the design of a new drug, the development of new materials for clean energy or addressing climate change can take place with high performance computing: as a matter of fact, in silico experimentation adds a powerful new dimension to knowledge discovery in all fields, alongside theory, physical experimentation and observation. With the integrated role of computation in theoretical and applied research, centers like CSCS and MGHPCC are a critical piece of infrastructure that will foster the world-leading innovation economy in their own countries through cooperative research, education and outreach activities.
In a stunning synchrony with CSCS, MGHPCC building as well was completed in 2012: it was the first university research data center to achieve the LEED Platinum Certification, the highest level awarded by the Green Building Council’s Leadership in Energy and Environmental Design Program, thanks to major investments in energy efficiency minimizing the amount of non-computing “overhead” energy used for cooling, lighting, and power distribution. In the same way CSCS achieved the Minergie TI-390 certification for efficient energy management.
Both computing centers bring overall energy savings and yield reduced carbon dioxide equivalent emissions with state of the art Power Usage Effectiveness (PUE). A common requirement to obtain energy efficient certifications is to reduce negative environmental impacts, with the attention to numerous details, including construction methods and materials, landscape and site design, and finally water conservation.
The group leaders and colleagues at my host institution have been an excellent source of deep knowledge in many respects in the HPC field, contributing tremendously to my professional development with their valuable experience in managing HPC support activities for scientific computing, in close interaction within a distributed network of partner institutions. The research computing support teams at the different founding institutions of MGHPCC administer advanced multiprocessor supercomputing systems for research computing and provide HPC application support for a vast academic community, with high-level technical and scientific user support, consulting, education, and training. The wide spectrum of disciplines that benefit from the infrastructure provided by MGHPCC and the partner institutions is reflected in the numerous applications and libraries supported, which include some of those available as well at CSCS in the fields of Quantum Chemistry, Molecular Modeling, Material Science, Data Analysis, Statistics and Bioinformatics.
There are significant overlaps between the activities of FAS Research Computing and CSCS UES unit, since the group where the applicant is employed at CSCS is involved in scientific computing support for the Swiss Academic community, with high-level technical and scientific user support, consulting, education, and training.
CSCS infrastructure supports as well numerous scientific applications and libraries for Quantum Chemistry, Molecular Modeling, Material Science, Data Analysis, Statistics and Life Science. In particular, the applicant’s background as a scientific application specialist fits well with the activities of the host group.
Furthermore, CSCS flagship system "Piz Daint" relies on a hybrid architecture with GPU Accelerators as well as the Odyssey Architecture that is managed by the FAS Research Computing group: Odyssey's scheduler is SLURM (Simple Linux Resource Manager), the same adopted on every computing system operated at CSCS. Therefore the interaction of the applicant with the host Institution will deal as well on the best practices to use effectively the SLURM scheduler in a work environment.
A range of other activities was covered in addition during my visit, like the usage of tools for automated regression testing and for building scientific applications. I also had the opportunity to visit Swissnex Boston, the Swiss honorary consulate in Cambridge, where I could interact fruitfully with the nice and welcoming staff. Furthermore, the extremely lively environment at Harvard University and at the surrounding institutions in the Boston area has been an excellent occasion to live and work in one of the most exciting places of the world.
In conclusion, the sabbatical has been for me a wonderful opportunity to interact successfully with experts and nice colleagues and experience the everyday working environment in one of the most advanced scientific institutions on earth: I am therefore very thankful to ETH and CSCS for the invaluable support they granted".