IT827
Course Name:
High Performance Computing (IT827)
Programme:
M.Tech (IT)
Credits (L-T-P):
(3-0-2) 4
Content:
High performance computing architectures, Fundamentals of Superscalar processors, Vector processors and Accelerators architecture – GPGPU, Xeon-Phis and FPGAs. Programming for HPC clusters – OpenMP and MPI programming. Programming for accelerators – OpenCL/CUDA/Xeon-PHIs. Domains for HPC (Microbiology, Engineering, Physics, Chemistry). HPC Benchmarks – LINPACK/HPL. Recent, relevant high-performance computing advances from literature from Supercomputing and other sources. Programming projects in the area of open source HPC applications. Introduction to Deep Learning frameworks like Caffe, Tensorflow. Implementation of HPC techniques in DNN frameworks incl. TPUs.
References:
Dennis Abts, John Kim, High Performance Datacenter Networks - Architectures, Algorithms, and Opportunities. Mark Hill/Margaret Martonosi (Eds.). Synthesis Lectures on Computer Architecture,Morgan and Claypool, 2011.
David B. Kirk, Wen-mei W. Hwu, Programming Massively Parallel Processors: A Hands-on Approach (Applications of GPU Computing Series) Elsevier-2014
David j. Kuck, “High Performance Computing”, Oxford Univ Press, 1996
Gary W. Sabot, “High Performance Computing”, Addison-Wesley, 1995
John L Hennessy, David A Patterson, Computer Architecture - A Quantitative Approach, 5th Edition,
Morgan Kaufmann, 2011.