WIR SCHAFFEN WISSEN HEUTE FR MORGEN Gerd Mann

WIR SCHAFFEN WISSEN – HEUTE FÜR MORGEN Gerd Mann : : Head IT and Head IT Department : : Paul Scherrer Institut Welcome @ PSI Joint JRA 2/ LEAPS WG 3/ Umbrella ID Meeting 13/05/2019

Overview Basel Germany Material sciences Radiochemistry Radiopharmacy Aarau/Bern Nanotechnology Hotlab Biology ESI Platform Zurich PSI East Swiss. FEL Particle physics Proton accelerator Energy research Neutron source Muon source Proton therapy PSI West Synchrotron Light Source Page 2

Our Mission Matter and materials Energy and environment Human health Development Construction Operation Knowledge & expertise Education Large research facilities Technology transfer Swiss and foreign users from academia and industry more that 2400 external users/year (39 beamports) Page 3

Key Figures PSI funds (global budget) External funding 280 MCHF 110 MCHF Staff (heads) 2100 • Externally financed 750 • Doctoral students 320 • Apprentices 100 External users: people / visits Number of scientific publications 2300 / 4800 per year 1400 (13 % of which high impact) per year PSI employees with teaching duties at both ETH and universities Patient visits (proton therapy treatment) 100 5900 per year Page 4

Challenge –Data Volume SLS and SLS 2. 0 250 • • 200 1 st ing Overvi e fore cast w plan n Roll IT requirements have not been a strategic factor in the past IT capabilities will be a key factor for the success of SLS 2. 0 211, 3 ed 150 100 84, 2 50 2, 0 0 SLS - 2018 SLS 2. 0 - 2024/2025 SLS 2. 0 - 2029 Raw and derived data (PByte) Seite 5

Work Break Down Structure for SLS 2. 0 Data Management & Data Processing Demand Management Data Readout and Online Processing Define Forecast Structure and Process Eiger Detector Rolling Forecast 2019 -2023 (Daa. S) Jungfrau 16 M Detector … Rolling Forecast 2024 -2029 (SLS 2. 0, …) … Data Reduction & Software Adaptation Architecture Design & Assessment & Implementation Evaluate Opportunities for Data Reduction Proof of Concept Remote Data Processing Implement Data Reduction Define Architecture(s) (Online/Offline Diffraction , SAXS, … (Veto/Compression) Adapt Software to Target Architecture Local/Remote CPU/GPU) Assess Architectures (Performance, Scalability, Cost) Operate and Renovate Operate Services Finalize Swiss. FEL Implement ATHOS Renovate Services (Life Cycle Mngmt, Scaling) Integrate Existing Facilities Implement Target Architecture Page 6

Data Processing – Remote Data Processing • PSI Goals - Proof of concept/pilot for the use of elastic HPC resources for - Online feedback exceeding PSI's local compute capacity - Integration of remote computational resources during a running experiment by providing fast feedback analysis and interactivity. This will be realized for ptychography experiments at PSI's c. SAXS beamline where the most resource consuming step of the analysis chain will be run on the CSCS resources. - Offline (re)-analysis of data that are already archived at CSCS - Serial crystallography data taken at Swiss. FEL (~100 TB) - Ptychographic reconstruction of a volume intensive experiment for the c. SAXS beamline

Wir schaffen Wissen – heute für morgen My thanks go to • Name Page 8