Conveners
Computations with Hybrid Systems (CPU, GPU, coprocessors)
- Dmitry Podgainy (JINR)
-
Nikolai Mester (Intel)29/09/2017, 11:20Sectional
-
Dr Dmitry Podgainy (JINR), Dr Oksana Streltsova (JINR)29/09/2017, 11:40SectionalWe apply several machine-learning (ML) algorithms for identification and separation of the neutron and gamma-ray signals coming from the DEMON (DEtecteur MOdulaire de Neutrons) detector. The ML-predictions have been contrasted with the results obtained within a standard method based on an integral-area scheme. In the situations where the standard method fails a properly trained ML-algorithm...Go to contribution page
-
Prof. Viacheslav Samarin (Joint Institute for Nuclear Research, Flerov Laboratory of Nuclear Reactions)29/09/2017, 11:55Computations with Hybrid Systems (CPU, GPU, coprocessors)SectionalThe modern parallel computing solutions were used to speed up the calculations by Feynman’s continual integrals method. The algorithm was implemented in C++ programming language. Calculations using NVIDIA CUDA technology were performed on the NVIDIA Tesla K40 accelerator installed within the heterogeneous cluster of the Laboratory of Information Technologies, Joint Institute for Nuclear...Go to contribution page
-
Sergey Belov (Joint Institute for Nuclear Research)29/09/2017, 12:10SectionalInteraction of labour market and educational system is a complex process, with many parties involved (government, universities, employers, individuals, etc.). Both horizontal and vertical mismatch between skills and qualifications from one side and market’s requirements from another are still widely observed in both developing and developed countries. To discover both qualitative and...Go to contribution page
-
Ms Victoria Tokareva (JINR)29/09/2017, 12:25Computations with Hybrid Systems (CPU, GPU, coprocessors)SectionalThe partial wave analysis at the BES-III experiment is being done event-by-event using the maximum likelihood estimation, with the typical statistics of the order of 10 billion J/ψ events per year, resulting in huge computation times. On the other hand, the event-by-event analysis can be naturally parallelized. We developed the parallel cross-platform software architecture that can run...Go to contribution page