The experiment name is the name of a directory that is created by the Collector to store the experiment data. The experiment name must end in .er to be accepted by the Collector. By default the experiment is created in the current directory.
If you do not explicitly name an experiment, a default name is constructed in the form stem.n.er. If you give an experiment group name, stem is the stem of the experiment group name (the part preceding .erg), otherwise it is test. The number n is a non-negative integer. For MPI programs n is the MPI rank. For other programs, the default value of n is 1.
If you give an experiment name in the form stem.n.er and the name is already in use, the Collector increments n until an unused name is found. Consequently, if you use the format stem.n.er for your experiment name, the Collector automatically increments the integer n by one in the names of subsequent experiments--for example, mytest.1.er is followed by mytest.2.er, mytest.3.er, and so on.
If you give an experiment name in the form stem.er and the name is already in use, the Collector prints an error message and does not accept the name.
Experiments on descendant processes are stored inside the experiment for the founder or parent process. The name stem for a descendant is constructed from the name stem of its creator by adding an underscore, a letter and a number. The letter is x if the descendant was created by a call to exec and f if the descendant was created by a call to fork. The number is the index of the fork or exec in the parent process. For example, an experiment recorded for an exec in the third fork of the parent process would have the name _f3_x1.er.
See also | |
---|---|
Experiment Groups MPI Programs |