Human Media Multicore Computing Lecture 1 : Course Overview Bong-Soo Sohn Associate Professor School of Computer Science and Engineering Chung-Ang University

Course Description Almost all recent PCs and smart phones/tablets are equipped with multicore CPUs Manycore GPUs are often adopted. Originally designed for 3D games Can be applied to general purpose (computing-intensive) applications Current trends in CPU/GPU technology indicate that the numbers of cores are expected to increase, rather than clock speeds.

Course Description parallel processing techniques utilizing multicore/manycore hardware become increasingly important for performance enhancement In this course, we will mainly focus on parallel programming principles and practice that utilize multicore/manycore computing system especially for human media applications

Course Objectives Understand multicore computing hardware Understand principles of parallel programming Practice actual parallel programming on real multicore/manycore computing systems with various libraries (pthread, OpenMP, CUDA, …) Conduct project on multicore computing for human media applications

Reference Books No Required Textbook. Reference Books “The Art of Multiprocessor Programming”, Maurice Herlihy, Nir Shavit “Programming Massively Parallel Processors: A Hands-on Approach”, David Kirk, Wen-mei Hwu Some of online materials will be used.

Evaluation Midterm Exam : 35% Final Exam : 35% Assignments + Projects : 25% Class Participation and Attendance : 5% You may get F grade if you miss ¼ of the whole classes

Instructor Information Name : Bong-Soo Sohn Office : 208-523-2 Email : bongbong@cau.ac.kr ( the best way to reach me) Tel : 820-5843 Office hour : Thursday 2pm-3pm, or individual appointment Please feel free to contact me for any of class issues.

Announcement Class Website Programming Assignments http://cau.ac.kr/~bongbong/multicore16 Class information such as lecture notes is accessible through this website We will also use e-Class for the submission of homework and project results. Programming Assignments I encourage you to study and discuss together for doing programming assignments. However, you must do programming YOURSELF. You must not share any of source code with other students. Academic dishonesty (e.g. cheating, plagiarism, and etc.) is seriously prohibited. Heavy penalty may be imposed for that.

Course Format Lecture Programming Projects powerpoint Slides Actual parallel programming on multicore systems.

Schedule 1 Course Introduction 2 Introduction to Multicore Computing 3 Mutual Exclusion 4 Concurrent Objects 5 Shared Memory 6 Synchronization 7 pthread (POSIX thread) 8 Midterm Exam

Schedule 9 pthread practice 10 threads on MS Windows OS 11 OpenMP 12 Manycore GPU Architecture and Programming 13 GPGPU : CUDA Parallel Programming 14 15 Student Presentation (result of team project) 16 Final Exam

Remarks This course assumes that students have basic knowledge of C and JAVA grammars and programming ability. We will have significant amount of programming assignments. If your programming ability is not good enough, it may be difficulty to follow the class.