- Parallel Computing overview (Chapters 1-2)
- Webpage with several MPI tutorials

- More MPI Intro (Ch 3.) Note: Webpage with several MPI tutorials

- Review of Unix and access to the IDI PC cluster.

- MPI Intro continued -- Ch 3-4. Note: Webpage with several MPI tutorials (MPI example: Parallel Numerical Integration/Trapezoidal)

Please note that this tutorial uses the C Shell, while the default shell at NTNU is the bash shell (GNU Bourne-Again SHell). To change to C shell, just type

csh

when you've logged in.

Section 3 introduces the a2ps command used for printing. To use it you need to know the name of the printer. Most IDI owned machines are configured to use the common printhost. For a list of printers, please see http://printhost.idi.ntnu.no/printers

Also note that the IDI version of a2ps has a few "quirks", so it's best to use it in piped mode. For example, to print to the printer on our floor, I use the commands

echo "Hei Hopp" | a2ps | lpr -Pitv443b

or

a2ps Makefile greetings.c greetings.pbs | lpr -Pitv443b

Today (Friday) at 13:15, Professor Jan Telle from The University of Bergen will be giving two talks in IT-emner. We meet in Kjel 3. Everyone is welcome.

Abstracts:

An Introduction to Computer Algorithms

Algorithms and their design and analysis are at the heart of information technology. Significant advances in algorithms are a crucial factor in the data processing underlying such technological developments as internet search engines, graphics applications and bioinformatics. It is a common misconception that as computers become faster the importance of the time complexity of the underlying algorithms becomes less significant. In fact, the opposite is true: Consider the size of the largest problem one can solve in a given amount of time. When the speed of the computer is increased the factor size(fast)/size(slow) between what an asymptotically fast algorithm and what a slow algorithm can solve will increase.

Graph Algorithms and Their Applications >p> Many real-world problems are modelled by a binary relation and solved by techniques from the field of graph algorithms. An important goal in the field of graph algorithms is to find large classes of graphs for which large classes of problems are solvable efficiently with practical algorithms. The arguably strongest results in this direction are achieved by the class of graphs of bounded treewidth, containing for example the control-flow graphs of structured programs. We will show how the compiler optimization problem of register allocation can be solved by this connection to bounded treewidth graphs.

Some material will also be taken from Goedecker & Hoise:
*Performance Optimization of Numerical Intensive Codes,* SIAM 2001,
and other handouts as indicated above.

Re. Ch 15, I will cover PETsc in more detail than in text. See http://www-fp.mcs.anl.gov/petsc/ for details

This page is maintained by : elster -at- idi.ntnu.no

Comments welcome. All above notes and assignments for TDT 4200 are COPYRIGHTed by Dr. Elster and/or the author(s). Any copying or further publication without the authors' consent is a COPYRIGHT violation.