Lectures with hand-on workshop will be given by IBM research scientist Dr. Leopold Grinberg with assistant Yu-Hang Tang from Applied Mathematics, Brown University.
Audience: CSRC faculty and researchers with interest in high performance scientific computing.
Registration: Please send your registration (your name and division) to Ms. Ying Fan, fanying@csrc.ac.cn
Objective of school: The four days summer school will focus on fundamentals of parallel computing on CPUs and GPUs.
Daily schedule:
8:30am-10:00am, 1st class
10:00am-10:30am, Coffee tea break
10:30am-12:00pm, 2nd Class
12:00pm-2:00pm, lunch break
2:00pm-3:30pm, 1st class
3:30pm-4:00pm, coffee tea break
4:00pm-5:00pm, 2nd Class
6:00pm, dinner
Lecture outline:
Day 1 (June 23, 2015)
Overview of the major components of compute nodes(CPUs, accelerators, memory, hierarchy, caches etc.)
Roof line model and its use in computational kernel analysis
Introduction into shared and distributed memory models.
Fine grain and coarse grain parallelism.
OpenMP: hands-on tutorials.
Day 2 (June 24, 2015)
Parallel computing: distributed memory model.
Message Passing Interface(MPI)
MPI: hands-on
Hybrid (distributed and shared memory) programming model.
MPI+OpenMP: hands-on
Day 3 (June 25, 2015)
Overview of GPU architecture
CUDA
Use of libraries (cuBLAS, cuSPARSE, THRUST)
Programming with CUDA: hands-on
Day 4 (June 26, 2015)
GPU memory hierarchy
GPU kernel optimization
Programming with CUDA: hands-on
Concluding remarks
Bio-sketch of Lecturers:
Dr. Leopold Grinberg: Dr. Grinberg is a research scientist at IBM research since 2013, his area of expertise is massively parallel solvers including molecular dynamics, CFD, seismic, and more. He focuses on algorithms, solver optimization, improving node performance, scaling (MPI, OpenMP), IO and hybrid CPU-GPU computing.
Education • Ph.D. Applied Mathematics, Brown University, USA, 2009.
Research Interests • Parallel computing with MPI, OpenMP (and hybrid), • Scaling (strong) applications to O(100K) CPU cores • Design and programming solvers using multilayer task and data parallelism, intrinsics,• Languages: C/C++, CUDA, Matlab, Fortran • Platforms: IBM BlueGene, IBM Power, CRAY, Linux Cluster. • High-order spectral/hp methods • Computational Fluid Dynamics • Linear solvers • Biomedical modeling • Cardiovascular Flows • Multiscale flow simulations.
Personal linkedin page:
https://www.linkedin.com/pub/leopold-grinberg/12/66b/651
Yu-Hang Tang: Mr. Tang is a Ph.D. candidate with the Division of Applied Mathematics at Brown University. His primary research interests focus on High Performance Computing and concurrent multiscale coupling with applications in modelling soft matter systems and physiological fluids. He is the author of several open-source software packages, including the LAMMPS USERMESO GPU-acclerated package for Dissipative Particle Dynamics (DPD) and Smoothed Particle Hydrodynamics (SPH) simulations, as well as the Multiscale Universal Interface library for coupling standalone solvers to perform multiscale simulations. Those software were used to carry out large-scale in silico investigation of amphiphilic polymer self-assembly using billions of particles. He is a reviewer of the Journal of Computational Science and a NDA member with NVIDIA.
Personal linkedin page:
http://www.dam.brown.edu/dpd/doku.php/yu-hang_tang
