Manage the base data for tuning the performance of parallel programs by profiling.
Detect performance issues and bottlenecks caused, for example, by inefficient programming, memory accesses, I/O operations, cache-misses, page-faults, and parallelization overheads.
Assess how different views of the profiling data (e.g. timeline graphs and communication matrices to illustrate the traffic between processes) can give insights into the runtime behavior of the program.
Use performance analysis tools like ScoreP, and Scalasca.