ECE 563 Programming Parallel Machines Spring 2017
Course materials, etc.
- The syllabus is now available. It was last updated on January 8, 2017.
- Homework assignments can be found at hw.html.
Last updated April 10, 2017, 1:01PM.
- Slides used in class can be found at slides.html.
Last updated February 27, 2017, 4:30AM.
- Papers and other materials can be found at other.html.
Last updated January 5, 2015 1:44PM
- Information on tests can be found at tests.html.
Last updated April 19, 2017 at 4:30PM
- Information on the projects can be found at projects.html. Last updated February 10, 2017 12:49PM and subject to change.
Other useful information
Shared memory -- how to compile and run OpenMP programs
- In general (e.g., qstruct and the wabash machines):
- In general, login into the machine you will run the job on
- To compile the program gcc -fopenmp filename.c
- To run the program a.out
- Note that most standard C compiler flags can be used, e.g., to have an output file named something other than a.out, etc.
- An example of running an OpenMP program on qstruct01.
- On the Scholar machines:
- Scholar should be used for timing runs.
- You do not yet have accounts on Wabash.
Shared memory -- OpenMP, Pthreads info
We will have three machines that MPI jobs can run on. They are:
- The scholar cluster: This queue gains you access to some of the nodes on Purdue's Carter cluster. Detailed instructions on how to run jobs on this machine can be found at
scholar guide. A quick reference card for running jobs can be found at scholar quick reference. Instructions and scripts provided by a student last semester can be found at scholar cookbook.
The scholar queue should be used for runs that you do timings on as it will minimize the interference of other jobs with your job. Debugging may be best done on the wabash and qstruct machines.
- wabash: this is my research machine. I will make some nodes available. If my graduate students need to use it for their research I'll let you know and you should not use it. MPI jobs should be run in the same way they are run on the qstruct cluster. These machines are configured similar to nodes on Purdue's rice cluster.
- qstruct: Like wabash, this is a machine that jobs run when submitted. If other jobs are running on the node you will run with them, competing for resources. Instructions on how to run jobs on the qstruct machines can be found here. This was updated on April 19, 2014.
- Instructions on how to not have to enter your password a gazillion
times every time you run an mpi program is available
here. These are slightly different than what I had before and show an actual session running these.
Your project will be much easier to do on a Linux system, and the quatro, qstruct and nile machines all run Linux. >A general Linux Tutorial is here and it describes various commands, etc.
Last updated March 3, 2017 7:33