The Performance Analyzer tool features the following other enhancements.
You can now specify a path in which to store Performance Analyzer's user settings by using the analyzer command's –u or –-userdir argument.
Comparing and aggregating experiments is now easier to do. You can choose between absolute values or deltas to show the change in metric values when comparing experiments.
Aggregation of performance metrics by shared library or Java class is greatly improved, and the dialog supporting it has been renamed from "Show/Hide/API-only" to "Set Library and Class Visibility
The Print option was replaced by Export to a file and supports more outputs. In some views, you can now export to an ASCII text table, a delimiter-separated list, or an HTML table. All exports are to the file system on the machine on which the Performance Analyzer is running. If you are running in remote mode, you export the data to a file on the remote machine.
The names of the metrics for clock profiling, MPI tracing, and heap tracing have changed. For details, see the collect (1) man page.
Performance Analyzer shows Java thread names and thread groups, and you can use them for filtering.
Performance Analyzer now supports attaching to a running process and profiling it. Choose File > Profile Running Process or click Profile Running Process on the Welcome screen. Previously you could only attach to a running process by using the collect command or dbx collector command.
You can now profile the Oracle Solaris kernel from Performance Analyzer. Choose File > Profile Kernel or click Profile Kernel on the Welcome screen. Previously you could only profile the kernel by using the er_kernel command.
Performance Analyzer has a new way to save settings for persistence. When you exit Performance Analyzer, most settings such as metrics and views are persistent, so that the next time you open the same experiment it is displayed the same way as when you last closed it. You can save selected settings in a configuration file and apply the configuration to the same or different experiments when you open them from the Open Experiment dialog box. Performance Analyzer can also save settings into a .er.rc file to be read by er_print.