CodeXL User GuideThis configuration is intended for advanced users who know what hardware events or combinations of events are important for specific analysis.
It allows a combination of hardware events, timer samples, and instruction-based sampling events. At least one event must be monitored for CodeXL to make the profiling.
Use of a smaller sampling period increases data collection overhead. Since data collection must be performed on the same platform as the test workload, more frequent sampling increases the intrusiveness of event-based profiling, and the sampling process adversely affects shared hardware resources such as instruction and data caches, translation lookaside buffers, and branch history tables. Extremely small sampling periods also can cause system instability. Start off conservatively and slowly decrease the sampling period for an event until the appropriate volume of samples is generated.
Warning You can demand too much sampling, causing the system to hang or crash. If this occurs, the easiest solution is to increase the sample intervals for the most popular events.
A factor when choosing the sampling period for an event is the workload behavior. Some workloads are CPU-intensive; other workloads are memory-intensive. Some workloads can be CPU-intensive and require high memory bandwidth to stream data into the CPU. For example, a CPU-intensive application that does not accesses memory very often causes relatively few data-cache miss events. The characteristics of the workload can even vary by phase, where the phase setting up a computation has a different behavior from the computation phase itself. Thus, the workload behavior determines the frequency of certain kinds of events, requiring changes to the sample period.
The available views for the profile data depend on the events selected. The events are selected through the _Ref371333729 Dialog.