For first tests of your code with few processes you should use the PCs in our computer laboratories. Please use only the machines in the student lab. If you measure your algorithm's speedup, you have to take into account the single nodes' performance. To find out about the speed of a node, log in on the node an type
cat /proc/cpuinfoMachine File:
To run the parallel version of your code you have to give MPI a file with all machines that can be used. MPI assigns the processes to the machines in the order given in the machine file. To list one node multiple times is allowed and makes MPI run multiple jobs on one machine. For the required format of the machine file see the MPI documentation or the example machine file.
email@example.com, Mehl, 14.01.2005