analyzer - Graphical tool for analyzing a program performance experiment
analyzer [-j|--jdkhome jvm-path] [-J jvm-options]
[-f|--fontsize size] [-u|--userdir dir_path]
[-v|--verbose] [experiment-name]
analyzer [-j|--jdkhome jvm-path] [-J jvm-options]
[-f|--fontsize size] [-u|--userdir dir_path]
[-v|--verbose] -c [base-group] [compare-group]
analyzer -V|-version
analyzer -?|-h|--help
analyzer [-f|--fontsize size] [-u|--userdir dir_path]
[-v|--verbose] target [target-arguments]
The analyzer command starts Performance Analyzer, which is a Java-based graphical data-analysis tool that you can use to analyze performance data that is collected from target programs. The data is collected by the Collector using the Performance Analyzer's Profile Application dialog box, the collect command, or the collector commands in dbx. The code coverage tool uncover can also invoke the Collector.
The Collector gathers performance information to create an experiment during the execution of a process. Performance Analyzer reads in such experiments, analyzes the data, and displays the data in tabular and graphical displays. A command-line version of Performance Analyzer is available as the er_print utility.
When Performance Analyzer is invoked on more than one experiment or experiment group, it aggregates data from the experiments. You can change the mode to compare experiments. For more information, see the Comparing Experiments section below.
Specify the path to the Java virtual machine (JVM) software for running Performance Analyzer. The default path is taken first by examining environment variables for a path to the JVM, in the order JDK_HOME, and then JAVA_PATH. If neither environment variable is set, the version found on your PATH is used. If none is found, /usr/java/bin/java is tried.
Specify JVM software options. Multiple -J arguments can be supplied. Note that there is no space between the -J flag and the jvm-option. Examples:
analyzer -J-d64 -- run the 64-bit analyzer (valid for Java 7 only) analyzer -J-Xmx2G -- run with maximum JVM memory of 2 GB (Default, 1 GB) analyzer -J-d64 -J-Xmx8G -- run the 64-bit analyzer with maximum JVM memory of 8 GB (-J-d64 valid for Java 7 only)
Specify compare mode. When specified, two experiments or experiment groups must be provided as arguments. Performance Analyzer opens the experiment or experiment groups in comparison mode with the first experiment or group as the base group and the second as the comparison group.
Each group can be a single experiment, or a group containing multiple experiments. If you want to include more than one experiment in a compare group, you must create an experiment-group file to use as a single argument to analyzer.
Specify the font size to be used in Performance Analyzer.
Specify the path to the user directory in which to store user settings.
Print version information and Java runtime arguments before starting.
Print version information and exit.
Print usage information and exit.
Specify an experiment to open. To specify more than one experiment, provide a space-separated list of experiment names, experiment group names, or both.
Specify a target program to be profiled along with any arguments for the program.
To start Performance Analyzer, type the following command on the command line:
analyzer [experiment-name]
If you do not specify experiment names when invoking Performance Analyzer, the initial view is the Welcome screen. You can open experiments from the Welcome screen or by using Performance Analyzer's File menu and toolbar buttons to open, compare, or aggregate experiments or experiment groups.
The optional experiment-name command argument is the name of an experiment, or a space-separated list of experiment names, experiment group names, or both. Experiments recorded on any supported architecture, can be displayed by Performance Analyzer running on the same or any other supported architecture.
Multiple experiments or experiment groups can be specified on the command line. For information about experiment groups, see Creating Experiment Groups section below.
If you specify an experiment that has descendant experiments inside it, all descendant experiments are automatically loaded and their data is displayed.
You can preview an experiment or experiment group for loading by single-clicking on its name in the Open Experiment dialog box.
You can start Performance Analyzer to profile an application from the command line as follows:
analyzer target [target-arguments]
Performance Analyzer starts up with the Profile Application dialog box showing the named target and its arguments, and settings for collecting an experiment. For more information, see Recording Experiments below.
To create an experiment group, create a plain text file whose first line is as follows:
#analyzer experiment group
Then add the names of the experiments on subsequent lines. The file extension of the experiment group text file must be .erg. You can use experiment groups also to load only specific descendant experiments if you want to isolate their data away from their founder experiment.
When you invoke Performance Analyzer on more than one experiment or experiment group, it normally aggregates the data from all the experiments. If you invoke Performance Analyzer with the -c flag, Performance Analyzer opens experiments in comparison mode.
You can also compare experiments from the Welcome screen, from the File menu, or from the Compare status area on the status bar.
In comparison mode, the Functions view shows separate columns of metrics for each experiment or group to enable you to compare the data.
Comparison style can be set from the Compare panel in Performance Analyzer or in the Settings dialog's Formats tab. By default, experiment comparison shows the absolute values of the base experiment and compared experiments. You can specify comparison style "Deltas" to show the metrics for the compared experiment as a + or - value relative to the base experiment. You can specify comparison style "Ratios" to show the compared experiment metrics as a ratio relative to the base experiment. Your selection is saved in your configuration settings for the next time you compare experiments.
Comparing experiments works in most data views except Call Tree, Races, Deadlocks, Heap, and I/O Activity.
The Source and Disassembly Views show a dual pane when comparing experiments. The Timeline shows separated data from the experiments being compared. The Selection Details window shows data from the base experiment only.
When Performance Analyzer is invoked with a target name and target arguments, it starts with the Profile Application dialog box open, allowing you to record an experiment on the named target when you click Run in the dialog. You can also record a new experiment by clicking Profile Application in the Welcome page, or clicking the Profile Application button on the toolbar, or choosing Profile Application from the File menu.
Press F1 in the Profile Application dialog box to see the Help for more information.
Note that the Profile Application dialog fields correspond to the options available in the collect command, as described in the collect(1) man page.
The Preview Command button at the bottom of the dialog box enables you to see the collect command that would be used when you click the Run button.
Performance Analyzer reads settings from a configuration file.
You can control the configuration settings from the Settings dialog box. To open this dialog box, click on the Settings button in the toolbar or choose Settings from the Tools menu. The Settings dialog box enables you to specify settings such as metrics and default data views.
You must click OK or Apply to apply your changes to the current session. Your settings are automatically saved to a configuration file in the experiment when you exit Performance Analyzer. The Settings dialog box has an Export button that you can use to save some or all of the settings in your configuration to other locations so you can share it with other experiments or users. When you open an experiment from the Open Experiment dialog, Performance Analyzer searches default locations for available setting configurations, and offers you the choices in the dialog. You can also import settings.
You can choose Tools > Export Settings to .er.rc file, in order to apply relevant settings to the .er.rc file which is used by the er_print utility.
A few of the possible directives are processed by Performance Analyzer from .er.rc files, including en_desc {on| off}. to control whether or not descendant experiments are selected and read when the founder experiment is read. Most other directives are ignored by Performance Analyzer, although they are processed by er_print.
The files also specify a path for C++ name demangling for other compilers.
When loading experiments with many descendants, you can set the SP_ANALYZER_DISCARD_TINY_EXPERIMENTS environment variable to control whether to ignore processes that had very short profile durations or had no profiling data. The following values are allowed:
-1 – Ignore processes without any profiling data. This is the default setting.
0 – Load all processes, regardless of duration or profiling data.
1 (or larger) – Set the minimum threshold in milliseconds for profile durations. Below this threshold, process data will be ignored.
For example, setting SP_ANALYZER_DISCARD_TINY_EXPERIMENTS=10 will ignore processes that have fewer than 10 milliseconds of profile data.
Performance Analyzer can run on a local system and connect to remote systems where the Oracle Developer Studio software is installed. You can then profile applications and read experiments located on the remote system.
A subset of the complete Performance Analyzer, called the Remote Analyzer, is available for local installation even on operating systems such as Windows or MacOS, which are not supported platforms for Oracle Developer Studio.
The Remote Analyzer is distributed as a tar file, RemoteAnalyzer.tar, in the lib/analyzer directory of the installed product. To install Remote Analyzer, copy the tar file to your local system and unpack it to create a subdirectory, RemoteAnalyzer.
The RemoteAnalyzer directory contains scripts for running the Oracle Developer Studio Performance Analyzer on systems that are not supported by the Oracle Developer Studio tools, or systems that are supported but do not have the tools installed. The directory also contains a lib directory with the necessary components for Performance Analyzer to execute.
There are four scripts, one each for Windows, MacOS, Oracle Solaris, or Linux. The scripts run Performance Analyzer on the local system and support connecting to a remote host. The remote host must have the Oracle Developer Studio tools installed, and the connection dialog requires you to enter the path to the tools on the remote machine.
Start Performance Analyzer on Windows by executing the AnalyzerWindows.bat file. You can type the command into a terminal window or double-click the file in the Windows Explorer. When you launch AnalyzerWindows.bat from Windows Explorer, it creates a terminal window and executes the command in that window.
When Performance Analyzer starts on Windows, it displays the Welcome screen. Many of the options are grayed out because they cannot be used locally on the Windows system. The main use of Performance Analyzer in this context is to connect to a remote host. The Welcome screen's documentation links work, but the links for profiling applications or reading or comparing experiments will not work until Performance Analyzer is connected to a remote host.
Start Performance Analyzer on MacOS by executing the AnalyzerMacOS.command file. You can type the AnalyzerMacOS.command into a terminal window or double-click the file in the Finder. When you launch AnalyzerMacOS.command from the Finder, it creates a terminal window and executes the command in that window.
When Performance Analyzer starts on MacOS, it displays the Welcome screen. Many of the options are grayed out because they cannot be used locally on the MacOS system. The main use of Performance Analyzer in this context is to connect to a remote host. The Welcome screen's documentation links work, but the links for profiling applications or reading or comparing experiments will not work until Performance Analyzer is connected to a remote host.
Start Performance Analyzer on Oracle Solaris or Linux by executing the corresponding Analyzer*.sh script in a terminal window.
When Performance Analyzer starts, it displays the Welcome screen. When the full Oracle Developer Studio suite is installed on the system where you are running Performance Analyzer, all of the Welcome screen options are enabled. When you run the Remote Analyzer on a system which does not have the full suite, many of the options are grayed out because they cannot be used without a Studio installation on the local system. The main use of Performance Analyzer in this context is to connect to a remote host. The Welcome screen's documentation links work, but the links for profiling applications or reading or comparing experiments will not work until Performance Analyzer is connected to a remote host.
You can connect to a remote host from the Welcome screen, the File menu, or from the Connected status area on the status bar. The Connect to Remote Host dialog enables you to specify the host and login information and then connect to the remote host. In the dialog, type the remote host name or select a host that you've used before, then type the name and password for a user account that can log in to that host, and type the path to the installation of Oracle Developer Studio on the remote host. Performance Analyzer remembers the last user name and path to installation that you used on each host.
Click Connect to log in to the host. Once you successfully log in, you again see the Welcome Screen, but this time all the options are enabled as if you were directly connected to the remote host.
Only used with Performance Analyzer's "Connect to Remote Host" option. If defined, disable periodic checking of the connection between Performance Analyzer and remote host. By default, a heartbeat packet is sent periodically from Performance Analyzer to the remote back-end. When enabled, the heartbeat packet can help keep some SSH connections from automatically closing due to inactivity.
Only used with Performance Analyzer's "Connect to Remote Host" option. Set interval for checking the connecting between Performance Analyzer and the remote host. The default is 2000 milliseconds.
Set this variable to /usr/bin/ssh to use /usr/bin/ssh for connecting Performance Analyzer to a remote host. If not set, use the internal jsch. For example:
SP_ANALYZER_REMOTE_SHELL="/usr/bin/ssh -u test-linux-nm1.pem"
On some Oracle Solaris systems, the X11 support which is needed is not installed. Performance Analyzer will report an error that it can't connect. The workaround is to run Performance Analyzer on a system with a display and X11 installed, and use the Remote Analyzer feature to connect to a remote host.
Sometimes Performance Analyzer will report that the GC Overhead limit was exceeded. The workaround is to use more memory than the default 1 GB. To run the analyzer with 2 GB, use:
analyzer -J-Xmx2G
To run a 64-bit version of Performance Analyzer with 8 GB of memory, use:
analyzer -J-d64 -J-Xmx8G
The option -J-d64 is only needed to run 64-bit when using Java 7 because Java 8 is 64-bit by default on Oracle Solaris and Linux.
Performance Analyzer will only work on experiments recorded with the current version of the tools. It will report an error for experiments recorded with any other version. You should use the version of Performance Analyzer from the release with which the experiment was recorded.
collect(1), collector(1), dbx(1), er_archive(1), er_cp(1), er_export(1), er_mv(1), er_print(1), er_rm(1), er_src(1), tha(1), libcollector(3)
Performance Analyzer manual
Performance Analyzer Tutorials