CSRC 2015 Summer School on HPC Parallel Computing

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


Slide L1.pdf

Slide-L2-roof_line_model.pdf

Slide L3-OpenMP.pdf

Slide-L4-MPI.pdf

Slide-L5-GPU.pdf

Slide-L6-GPU.pdf



CSRC 新闻 CSRC News CSRC Events CSRC Seminars CSRC Divisions 孙昌璞院士个人主页