The following figure shows metric sets for both core modules and memory modules. These metric sets are mutually exclusive and can be used in a combination based on the requirements of your design. Initially, it is recommended to use the heat_map metric for core modules and conflicts metric for memory modules. These settings are specified at run-time and therefore, you can run as many times as you would like with your preferences. However, many of these metrics sets are interconnected because some use group events and some use solely individual events. For example, the heat_map metric contains a group stalls event and two events for core execution, vector instructions and group core instruction using all data transfer instructions (for example, load/store, stream and cascade put/get, and lock acquire/release). To get a better view of which stall type(s) are prevalent, re-run with the stalls set. To better understand core execution, re-run with the execution set.
Figure 1. Core vs Memory Modules Metric Sets