Prostor: FAMNIT-1-MP2 at 16:00 – 18:00
Naslov: Basics of GPU programming using Cuda
Predavatelj: Prof. Nouredine Melab
Povzetek:
On the road to the exascale era, coprocessors including (GPU, Xeon Phi, etc.) are increasingly becoming key building blocks of High Performance Computing platforms. Besides their energy efficiency, these many-core devices boost the performance of multi-core processors through the combination of a larger number of processing cores, vector-SIMD processing and multi-threading. According to the last edition (November 2016) of the Top500 international ranking of supercomputers, GPU accelerators are the widely spread coprocessors which are successfully popularized by the easy-to-adopt programming models such as CUDA. The objective of this lecture is to give an introduction to GPU programming. The lecture starts with a reminder of parallel computing platforms and basics of parallel algorithm design and models. Then, the focus is put on the GPU programming using CUDA. The complex GPU hardware architecture is presented together with its associated parallel programming model. The basic features of CUDA are exposed and illustrated through examples. Some complexity analysis and performance results are reported.
Biografija:
Nouredine Melab is a full Professor at Université Lille 1, Inria Lille and CNRS CRIStAL labs. He has received a PhD in computer science from Université Lille 1 in 1997. He has been an Associate Professor at Université du Littoral – Côte d’Opale then Polytech’Lille before joining Université Lille 1. His research interest include combinatorial optimization and parallel (multi-core, many-core, cluster, etc.) computing. He has (co-)authored more than 150 international publications on parallel and/or distributed optimization including journal and conference papers and book chapters. He has supervised 15 PhD theses on parallel optimization. He is the scientific leader of supercomputing at Université Lille 1 since 2005. In Education, N. Melab is the co-head of the master of advanced scientific computing at Lille 1 since 2010. He is teaching mainly operations research and parallel and distributed computing.
Predavanje je organizirano v okviru PONEDELJKOVEGA SEMINARJA RAčUNALNIŠTVA IN INFORMATIKE Oddelkov za Informacijske znanosti in tehnologije UP FAMNIT in UP IAM in SYNERGY – H2020 TWINNING projekta.
Vabljeni!
