ENEE651/CMSC751: Parallel Algorithms, Spring 2020

Time and Location

• TuTh 2:00-3:15. IRB 1207

Instructor

• Dr. U. Vishkin, vishkin@umd.edu
• Office hours: Tue 3:30-4:30 (by appointment) at IRB 5216

TA

• Ananth Hari, ahari1@terpmail.umd.edu
• Office hours: Wed & Fri 11:00-12:00 at AVW 1301

Programming Assignments

Dry Homework

Course Goals

• Introduction to the theory of parallel algorithms. The course will highlight parallel algorithmic thinking for obtaining good speed-ups with respect to serial algorithms.
  The main objective of the class presentations and the written homework will be to study the main theory of parallel algorithms. This includes the design and analysis of parallel algorithms--primarily standard asymptotic analysis. The main objective of the programming assignments will be to reduce this knowledge to practice for improved understanding. One highlight will be achieving hard speedups for the more advanced parallel algorithms studied, demonstrating to students that they can do it even within the limited scope of course programming assignments. The programs will run on real parallel hardware whose design objectives included cost-effective support for the parallel algorithms theory studied.
• Most opportunities for performance growth in computing since 2005 have been tied to their exploitation of parallelism. Parallel computing has made big strides and is being used on an unprecedented scale within companies like Google and Facebook, for supercomputing applications and in the form of GPUs and multi-cores. The class will review some of the most exciting applied research issues facing computer science and engineering, as a result of these opportunities. In particular, it will include a close examination of the the extent to which the current ``Multi-Billion-transistor-per-chip'' era provides a way for building a truly general-purpose parallel computer system on-chip.

Of Special Interest

• 4-6 parallel programming assignments will be given.
• To be run on the Paraleap, the UMD XMT 64-processor computer.

Prerequisite Topics

• Basic knowledge and understanding of algorithms and data structures and computing systems.

Course readings:

• Class notes and research papers will be provided by the instructor. Please download and print out the first item (104 pages) under http://www.umiacs.umd.edu/users/vishkin/PUBLICATIONS/papers.html
  
  Reference books:
• J. JaJa, An Introduction to Parallel Algorithms, Addison Wesley, 1992.
• D.E. Culler and J.P. Singh, Parallel Computer Architecture, Morgan Kaufmann, 1999. The following quote appears under the title Potential Breakthroughs: ..."breakthrough may come from architecture" ... "that is, to truly design a machine that can look to the programmer like a PRAM".
• J.L. Hennessy and D.A. Patterson. Computer Architecture a Quantitative Approach, 6th Edition. Morgan-Kaufmann, San Francisco, 2017.

Some Core Topics:

• Introduction to Parallel Algorithms: Performance of Parallel Algorithms Including Work and Depth, Optimality Notions
• Basic Techniques: Balanced Trees, Pointer Jumping, Divide-and-Conquer, Partitioning, Pipelining, Accelerated Cascading, Breaking Symmetry
• Trees and Lists: List Ranking, The Euler Tour Techniques, Tree Connection, Evaluation of Arithmetic Expressions
• Searching, Merging and Sorting
• Graphs: Connected Components, Transitive Closure, Paths Problems
• Strings: Some string matching algorithms
• Introduction to Parallel Processing
• Introduction to state-of-the-art of parallel computing and discussion of the question: Can fine-grained large scale on-chip parallel computing be harnessed towards faster completion time of a single general-purpose computing task?

Grading

• 10% - Written homework.
• 25% - Programming projects. You must submit all programming projects to get a grade.
• 25% - 1st mid-term exam. Exam date: March 12, in class.
• 40% - 2nd mid-term exam. Exam date: May 5, in class. There will be no final.