26 04 im alumniV8
22 11 im fatiado face
22 11 im fatiado twitter
22 11 im fatiado youtube
22 11 im fatiado gmail
22 11 im fatiado brazil
22 11 im fatiado england
22 11 im fatiado spain

13 05 im DestaqueNoticiaPalestrante: Prof. Claude Tadonki, do Centre de Recherche en Informatique da École de Mines de Paris
Data: 13/05/2019 (segunda-feira)
Horário: 15:00 - 16:00
Local: Sala C-119

Abstract: Multi-core has become the standard architecture model for modern processors. As the main trend is to increase the number of cores on a single chip, the associated hierarchical memory system, which is concurrently shared by all available compute units, is becoming more complex and less predictable. Many common scientific and technical computation kernels suffer from a significant overhead from memory activities, which stands as a bottleneck and mostly bounds the overall performance. On top of this, any parallel application is subject to fundamental and conceptual aspects that moderate the expectations for a perfect scalability. In addition, hardware mechanisms incur some overhead in case of concurrent actions. Moreover, the effective scheduling of the tasks and the management of their interactions by the system also yield an extra time cost. All these aforementioned aspects affect the expected scalability and efficiency on multiprocessor systems. The aim of this talk is to provide a broad view of the main hardware and software hindering points related to parallel efficiency on manycore processors and tips to overcome them at the best.

Biography: Claude Tadonki (M) is a senior researcher and lecturer at the MINES ParisTech - PSL (Paris/France) since 2011. He holds a PhD and an HDR in computer science from University of Rennes and from Paris-Sud University respectively . After six years of cutting-edge research in operational research and theoretical computer science at the University of Geneva, he relocated to France to work for EMBL, University of Paris-Sud, LAL-CNRS and then MINES ParisTech. His main research topics included High Performance Computing, Parallel Computing, Operational Research, Matrix Computation, Combinatorial Algorithm and Complexity, Mathematical Programming, Scientific and Technical Programming, Automatic Code Transformations. Claude Tadonki has worked at several laboratories and universities, has initiated various scientific projects and national/international collaborations, has given significant number of CS courses in different contexts including industries. He is an active member of well established scientific corporations and reviewer of high-impact international journals and top-rank conferences. He has published numerous papers in journals and international conferences.He is very active in international collaborations and has co-organized several HPC conferences and forums.