2020-B-Nakajima Kengo

Advanced Course on Multi-Threaded Parallel Programming using OpenMP for Multicore/Manycore Systems

August 22, August 29, and September 5, 2020

Online-Course by Cisco Webex (Video recorded materials are also available)

Registration

Please register online before July 31, 2020: https://forms.gle/2ds6zmjgknQJmfZN6

(The number of participants is limited. You will receive notification if registered and enrolled successfully.)

Announcements

  • TBA

Course Handouts

  • TBA

Overview

In order to make full use of modern supercomputer systems with multicore/manycore architectures, hybrid parallel programming with message-passing and multithreading is essential. While MPI is widely used for message-passing, OpenMP for CPU and OpenACC for GPU are the most popular ways for multithreading on multicore/manycore clusters. In this 3-day course, we focus on optimization of single node performance using OpenMP for CPU. We “parallelize” a finite-volume method (FVM) code with Krylov iterative solvers for Poisson’s equation on Oakbridge-CX supercomputer (OBCX) at the University of Tokyo with 6.6 PF peak performance (https://www.cc.u-tokyo.ac.jp/en/supercomputer/obcx/system.php), which consists of the most recent CPU’s (Intel Xeon Platinum 8280 (Cascade Lake, CLX).

In this Summer School, the target application is a 3D FVM code for Poisson’s equation by ICCG Method (Conjugate Gradient (CG) iterative method with Incomplete Cholesky preconditioning), which is widely-used in practical applications. Because ICCG includes “data dependency”, where writing/reading data to/from memory could occur simultaneously, parallelization using OpenMP is not straight forward. We need certain kind of reordering in order to extract parallelism. In this 3-day course, lectures and exercise on the following issues will be provided:

  • Overview of Finite-Volume Method (FVM)

  • Kyrilov Iterative Method, Preconditioning

  • Implementation of the Program

  • Introduction to OpenMP

  • Reordering/Coloring Method

  • Parallel FVM by OpenMP

Schedule

August 22, 2020 (Sat 9:10-17:00)

09:10-10:00 Introduction

10:10-11:00 Finite Volume Method (FVM) (1/6)

11:10-12:00 Finite Volume Method (FVM) (2/6)

13:10-14:00 Finite Volume Method (FVM) (3/6)

14:10-15:00 Finite Volume Method (FVM) (4/6)

15:10-16:00 Finite Volume Method (FVM) (5/6)

16:10-17:00 Finite Volume Method (FVM) (6/6)

August 29, 2020 (Sat 9:10-17:00)

09:10-10:00 Introduction to OpenMP (1/3)

10:10-11:00 Login to OBCX System

11:10-12:00 Introduction to OpenMP (2/3)

13:10-14:00 Introduction to OpenMP (3/3)

14:10-15:00 Reordering (1/5)

15:10-16:00 Reordering (2/5)

16:10-17:00 Reordering (3/5)

September 5, 2020 (Sat 9:10-17:00)

09:10-10:00 Reordering (4/5)

10:10-11:00 Reordering (5/5)

11:10-12:00 Parallel FVM by OpenMP (1/5)

13:10-14:00 Parallel FVM by OpenMP (2/5)

14:10-15:00 Parallel FVM by OpenMP (3/5)

15:10-16:00 Parallel FVM by OpenMP (4/5)

16:10-17:00 Parallel FVM by OpenMP (5/5)

Prerequisites

  • Experiences in Unix/Linux (vi or emacs)

  • Experiences in Programming by C/C++/Fortran

  • Fundamental numerical algorithms (Gaussian Elimination, LU Factorization, Jacobi/Gauss-Seidel/SOR Iterative Solvers, Conjugate Gradient Method (CG))

  • Experiences in SSH Public Key Authentication Method (optional)

Preparation

Instructor

Prof. Kengo Nakajima

Information Technology Center, The University of Tokyo, Tokyo, Japan [Web]

主辦單位：科技部 國家理論科學研究中心數學組

協辦單位：東京大學情報基盤中心，科技部人工智慧技術暨全幅健康照護聯合研究中心，台灣工業與應用數學學會，臺灣大學數學系與應用數學科學研究所

主持人：王偉仲 (台灣大學應數所)，林文偉 (交通大學應數系)，舒宇宸 (成功大學數學系)，黃聰明 (台灣師範大學數學系)

聯絡人：Mr. Hsieh (02-3366-8819, medalab.contact@gmail.com), Medical Data Analytics Laboratory (MeDA Lab, http://meda.ai)