ECS 158, Fall 2017

Norm Matloff, matloff@cs.ucdavis.edu

Brief Course Overview

• Content of the course: Programming on multicore, clusters and GPUs. Industry-standard tools used, i.e. OpenMP, MPI, and CUDA, all of which we will use from C/C++. Some R will also be used. (Prior background in OpenMP/MPI/CUDA/R not needed.)
• Practical value of the course: Parallel computation is a Must in this era of Big Data. Many students have gotten jobs in industry specifically for their knowledge of ECS 158.
• Background needed: First and foremost, good programming skill. Parallel code is more challenging to write than ordinary code. In the quizzes you will be writing actual code, NOT answering questions about syntax etc. If you are not comfortable with writing intricate code from scratch without being given an outline to work from, this may not be the course for you.

  Also, basic understanding of computer systems, i.e. OS processes, memory structure etc., and basic understanding of matrix algebra (MAT 22A/67). The OS and matrix algebra material will be reviewed as part of the course.

• Exams: No midterms or final. (Term project in lieu of final.) Quizzes in discussion section almost every week. Special Group quiz on the last day of lecture. Quizzes will be taken (in class) on your laptop using https://github.com/matloff/omsi (to your great advantage).
• Textbook: My free, open-source book. Detailed reading is required.
• Group work: The programming assignments, group quiz and term project are all done in groups of 3 or 4 students.
• Grading: Quizzes 70%, programming assignments/term project 30%.
• My own background: Parallel computation is one of my research fields, and I have a published (non-open source) book in this area. See my bio for my general background.