Some classical performance analysis tools are also capable of acquiring and displaying performance metric data as it can be found on most parallel architectures today. This data is typically generated on a per process basis or for groups of processes, which makes it relatively easy to distribute the data among the workers. Similar to block entry/exit events, profiles and timelines can be calculated independently on the workers. In a subsequent step, the control process merges the condensed data and passes the results over to a client.