Benchmarking

A) datafile for mike

Command for running the script:

gnuplot -persist mike_script_all_proc.txt

This will generate a scaling graph for the Bench_Whisky_Carpet_48l:

or having the points connected for analysis of the expected output:


B) datafile for pelican

Command for running the script:

gnuplot -persist pelican_script_all_proc.txt

Here is the scaling graph for the Bench_Whisky_Carpet_48l:

or having the points connected - this shows the increase we expected:

The results for Pelican are not complete. There are still a few jobs for 64 and 96 processors running on the LSU supercomputer.

We also compiled Whisky_Carpet with -r option which made the runnings slower.

Here is the script and here is the datafile.


Results for one processor for different machines:

The command to run the script is the same with the previous ones, except for the name of the script.

gnuplot -persist whisky1proc.txt

The datafile can be seen here.