The purpose of this assignment is to get you engaged in the course and ready to do assignment #1. Assignment #0 is not worth any marks, but I do want a report from you handed in, and I will be giving them a grade of Unsatisfactory/Satisfactory/Excellent. I would like you take this assignment seriously. For each question, just follow the directions about what to do and what to report. If you had trouble with the question, please describe what the difficulty was.
Visit the course web site at http://www.sable.mcgill.ca/~hendren/621/index.html and the My Courses for COMP 621. On the MyCourses site, introduce yourself on the discussion list entitled ``Introduce yourself".
If you have already taken an introductory compiler course, then the answer for this question is just the name, and a short description of the course.
If you have not taken an introductory compiler course, then you can fill in the background you need for this course by doing some reading from any standard compiler textbook. To answer this question say what reading you did to prepare.
In this course you will be expected to be able to understand large frameworks written in Java, and if you plan to do a project with the VM implementation, you should be comfortable with C++. To answer this question, please summarize your experience with Java and C++.
If you don't have significant experience, especially with Java, then please do some background reading on Java, and write some practice programs in Java. If this is your situation, please summarize how you prepared for Java.
MATLAB is a very popular language that is challenging for compiler optimizations. MATLAB was made to be easy for Scientists/Engineers to use, but it is actually very hard to analyze and optimize. In this course we will use the McLAB analysis framework for MATLAB to practice with designing new compiler analyses. However, before you can understand how to build the analyses you need to understand the foundations of the MATLAB language.
Both Mathworks' MATLAB and an open source version, Octave, are installed on the SOCS machines in Trottier. You may also find them on the research machines in your graduate labs (if they are not there, please ask your research groups to install them, McGill has an institutional license to install on all university-owned computers).
If you are not familiar with MATLAB, you can refer to the following documents:
To experiment with MATLAB and to start to understand performance issues with MATLAB, you should write a short numerical program, one version using MATLAB vector notation and one version using explicit loops. For the explicit loops version, be sure to ``pre-allocate" any arrays you are defining in the body of the loop using the built-in zeros function. This is to prevent MATLAB from growing the array on each loop iteration.
You can use code you find online or in a book, but make sure you say where you got the code from, and make sure you understand the code thoroughly. You can also write your own code.
Check that your vector version and your loop version compute the same results.
Time the two versions for various problem sizes. Hint: you can use the commands tic and toc.
To answer this question, provide the two versions of your program, and
briefly describe the results of your timing, and your interpretation of
those results. You might have fun searching the web for various advice
about how to use vectors to speed up MATLAB, and seeing if that advice
is true or not. One relevant research paper can be found at:
http://dl.acm.org/citation.cfm?id=331972 .1
Hint for this question: MATLAB has a "JIT-accelerator" now, whereas before it was only an interpreter.
If you feel even more ambitious, try running MATLAB in interpreter-only mode and repeat the experiment. You can turn the JIT on and off using the MATLAB commands feature accel on and feature accel off. It is on by default. What did you find?
To be even more ambitious, try it also on Octave, which is an open source interpreter. What did you find?
Briefly describe your research interests and how COMP 621 fits with those interests.
This document was generated using the LaTeX2HTML translator Version 2008 (1.71)
Copyright © 1993, 1994, 1995, 1996,
Nikos Drakos,
Computer Based Learning Unit, University of Leeds.
Copyright © 1997, 1998, 1999,
Ross Moore,
Mathematics Department, Macquarie University, Sydney.
The command line arguments were:
latex2html -local_icons -split 0 a0.tex
The translation was initiated by Laurie HENDREN on 2015-09-04