PCAC_HELP.HLB  —  pcacui  Overview frame
  PCA collects performance information on your
  application and lets you view that information
  to observe the runtime characteristics of your
  application, such as:

  o  Where your application spends the most time

  o  What parts of your application are executed the most

  o  What parts of your application are compute-bound or
     I/O-bound

  o  What areas of your application are, or are not,
     executed by tests

  o  What the dynamic call relationships are among the
     routines in your application

  o  Where page faults occur

  o  Where you use system services

  With this information, PCA helps you pinpoint important
  qualities of your application, such as the location
  and nature of performance bottlenecks, or how effective
  your test coverage is.

1  –  ElaTimInfo frame

  Collecting elapsed time information gives you a broad
  measure of where your application is spending the most
  elapsed, or wall-clock, time. Elapsed time includes
  wait loops and other factors, both internal and
  external to your application, that affect how long
  it takes for your application to complete.

  To collect elapsed time information on your
  application:

  o  Compile and link your application with the PCA
     Collector

  o  Run your application. This causes the PCA Collector
     to take control of your application. PCA diplays the
     Collector main window on your screen.

  o  From the File pull-down menu, create a data file
     into which the Collector places performance data.

  o  Choose the Process Execution query category.

  o  Click on the elapsed time query.

  o  Click on Collect in the Collector main window.

  For information on displaying the results of the
  elapsed time data collection, you can refer to:

  o  The hardcopy manual, Guide to DIGITAL Performance
     and Coverage Analyzer for OpenVMS Systems

  o  The online help for the PCA Analyzer.

2  –  ProTimInfo frame

  Collecting process time information on your application
  tells you where your application is spending the most
  process, or CPU, time. Process time excludes external
  factors such as device wait loops and system load from
  the data sampling on your application. Thus you get a
  good indication of where your application has compute-
  bound performance bottlenecks.

  To collect process time information on your
  application:

  o  Compile and link your application with the PCA
     Collector

  o  Run your application. This causes the PCA Collector
     to take control of your application. PCA diplays the
     Collector main window on your screen.

  o  From the File pull-down menu, create a data file
     into which the Collector places performance data.

  o  Choose the Process Execution query category.

  o  Click on the process time query.

  o  Click on Collect in the Collector main window.

  For information on displaying the results of the
  process time query, refer to:

  o  The hardcopy manual, Guide to DIGITAL Performance
     and Coverage Analyzer for OpenVMS Systems

  o  The online help for the PCA Analyzer.

3  –  TesCovInfo frame

  Collecting test coverage information on your
  application can tell you how effectively and completely
  your tests exercise the codepaths in your application.
  Test coverage information can tell you which lines of
  code were, or were not, covered.

  To collect test coverage information on your
  application:

  o  Compile and link your application with the PCA
     Collector

  o  Run your application. This causes the PCA Collector
     to take control of your application. PCA diplays the
     Collector main window on your screen.

  o  From the File pull-down menu, create a data file
     into which the Collector places performance data.

  o  Choose the Coverage query category.

  o  Click on the query related to the coverage
     information you are interested in.

  o  Click on Collect in the Collector main window.

  For information on displaying the results of the
  coverage queries, refer to:

  o  The hardcopy manual, Guide to DIGITAL Performance
     and Coverage Analyzer for OpenVMS Systems

  o  The online help for the PCA Analyzer.

4  –  ExeCtsInfo frame

  Collecting execution counts information tells you
  what parts of your application are executed most when
  you place breakpoints at specific program locations.
  Execution counts information can be very useful for
  giving clues to the reason why a particular routine or
  algorithm is consuming a lot of time.

  To collect counting information on your application:

  o  Compile and link your application with the PCA
     Collector

  o  Run your application. This causes the PCA Collector
     to take control of your application. PCA diplays the
     Collector main window on your screen.

  o  From the File pull-down menu, create a data file
     into which the Collector places performance data.

  o  Choose the Execution Counts query category.

  o  Click on the query related to the exact counting
     information you want.

  o  Click on Collect in the Collector main window.

  For information on displaying the results of the
  execution counts query category, refer to:

  o  The hardcopy manual, Guide to DIGITAL Performance
     and Coverage Analyzer for OpenVMS Systems

  o  The online help for the PCA Analyzer.

5  –  IoOpsInfo frame

  Collecting I/O usage information tells you where your
  application is performing I/O operations, and gives
  detailed information on each I/O operation it records.
  Collecting I/O information can give you more detailed
  information than you can get from collecting system
  services information.

  To collect I/O usage information on your application:

  o  Compile and link your application with the PCA
     Collector

  o  Run your application. This causes the PCA Collector
     to take control of your application. PCA diplays the
     Collector main window on your screen.

  o  From the File pull-down menu, create a data file
     into which the Collector places performance data.

  o  Choose the I/O Usage query category.

  o  Click on the specific I/O query you are interested
     in getting information on.

  o  Click on Collect in the Collector main window.

  For information on displaying the results of the I/O
  Usage queries, refer to:

  o  The hardcopy manual, Guide to DIGITAL Performance
     and Coverage Analyzer for OpenVMS Systems

  o  The online help for the PCA Analyzer.

6  –  SysSerInfo frame

  Collecting system service usage information on your
  application is useful for determining where and your
  program calls each system service and what parts of
  your application use system services the most. For each
  system service call, PCA gathers the following data:

  o  The system service index

  o  The PCA address of the system service call

  o  The current process, or CPU, time

  To collect system service usage information on your
  application:

  o  Compile and link your application with the PCA
     Collector

  o  Run your application. This causes the PCA Collector
     to take control of your application. PCA diplays the
     Collector main window on your screen.

  o  From the File pull-down menu, create a data file
     into which the Collector places performance data.

  o  Choose the System Service Uages query category.

  o  Click on the specific system service usage query
     that you are interested in collecting information
     on.

  o  Click on Collect in the Collector main window.

  For information on displaying the results of the system
  service usage queries, refer to:

  o  The hardcopy manual, Guide to DIGITAL Performance
     and Coverage Analyzer for OpenVMS Systems

  o  The online help for the PCA Analyzer.

7  –  AdaTskInfo frame

  Collecting Ada Tasking information gives you the
  following tasking information on your Ada application:

  o  What tasks are performing context switching

  o  What Ada tasking names are used

  o  What Ada tasking types are used

  o  What Ada task priorities are used

  To collect Ada Tasking information on your application:

  o  Compile and link your application with the PCA
     Collector

  o  Run your application. This causes the PCA Collector
     to take control of your application. PCA diplays the
     Collector main window on your screen.

  o  From the File pull-down menu, create a data file
     into which the Collector places performance data.

  o  Choose the Ada Tasking query category.

  o  Click on the query related to the specific tasking
     information you are interested in collecting.

  o  Click on Collect in the Collector main window.

  For information on displaying the results of the Ada
  tasking queries, refer to:

  o  The hardcopy manual, Guide to DIGITAL Performance
     and Coverage Analyzer for OpenVMS Systems

  o  The online help for the PCA Analyzer.

8  –  RouCalInfo frame

  Collecting Dynamic Call information on your application
  tells you:

  o  Your application's call tree structure

  o  What routines call other routines

  o  Which routines make the most calls

  Unlike the static call stree structure provided by
  SCA, the dynamic call tree produced by PCA shows the
  runtime characteristics of the relationships among your
  application's routines during a given test run.

  To collect Dynamic Call information on your
  application:

  o  Compile and link your application with the PCA
     Collector

  o  Run your application. This causes the PCA Collector
     to take control of your application. PCA diplays the
     Collector main window on your screen.

  o  From the File pull-down menu, create a data file
     into which the Collector places performance data.

  o  Choose the Dynamic Call Information query category.

  o  Click on the specific query related to the
     information you are interested in collecting.

  o  Click on Collect in the Collector main window.

  For information on displaying the results of the
  Dynamic Call Information queries, refer to:

  o  The hardcopy manual, Guide to DIGITAL Performance
     and Coverage Analyzer for OpenVMS Systems

  o  The online help for the PCA Analyzer.

9  –  PagFltInfo frame

  Collecting Page Fault information on your application
  tells you where page faults are occurring and what
  addresses in your application are causing page faults.

  To collect Page Fault information on your application:

  o  Compile and link your application with the PCA
     Collector

  o  Run your application. This causes the PCA Collector
     to take control of your application. PCA diplays the
     Collector main window on your screen.

  o  From the File pull-down menu, create a data file
     into which the Collector places performance data.

  o  Choose the Page Faults query category.

  o  Click on the query related to the specific page
     fault information you are interesting in collecting.

  o  Click on Collect in the Collector main window.

  For information on displaying the results of the Page
  Faults queries, refer to:

  o  The hardcopy manual, Guide to DIGITAL Performance
     and Coverage Analyzer for OpenVMS Systems

  o  The online help for the PCA Analyzer.

10  –  DisPerInfo frame

  To display the results of the performance data you have
  collected on your application:

  o  Invoke the Analyzer from the main PCA startup window

  o  Select the query category you collected performance
     information on

  o  Select the specific query you chose during the
     collection run

  The Analyzer diplays performance data in one of the
  following types of charts:

  o  Histograms

  o  Tables

  o  Annotated source listings

  o  Call trees

  For more information in displaying performance results,
  refer to:

  o  The online help for the Analyzer

  o  The hardcopy manual, Guide to DIGITAL Performance
     and Coverage Analyzer for OpenVMS Systems

11  –  LnkColInfo frame

  To compile and link your application with the PCA
  Collector:

  1. Compile your sources with the "/DEBUG" qualifier;

     This puts all the necessary symbolic information PCA
     needs into the object file.

  2. The following steps specifies the PCA Collector as
     your program's "debugger":

      i  Link your program with the "/DEBUG" qualifier;

     ii  Define the logical name LIB$DEBUG as
         PCA$COLLECTOR;

  3. If PCAC$DECW$DISPLAY is defined ensure it is not set
     to " ";

  4. If PCAC$DECW$DISPLAY is undefined ensure
     PCA$DECW$DISPLAY is not set to " ";

  5. Define a valid display device:

     a.  Either:

          i Use the "SET DISPLAY/CREATE/NODE=node-name"
            command to define the display device;

     b.  Or:

          i Define the logical name DECW$DISPLAY as a
            valid display device;

  6. Run your program;

  7. When the Collector window appears on the screen, you
     have successfully invoked the Collector.

12  –  HowToColl frame

  To collect performance information on your application:

  o  Compile and link your application with the PCA
     Collector

  o  Issue the run command for your application. This
     causes the PCA Collector window to appear on your
     screen.

  o  In the Collector window, click on the File pull-down
     menu and choose New Datafile... to create a data
     file into which the Collector places performance
     data.

  o  Click on the Query Category button to see the
     Query Category menu, from which you can select the
     category of performance data you are interested in
     collecting.

  o  Click on the specific query category you are
     interested in. This causes the specific queries
     for that query category to be displayed in the
     query pane at the bottom half of the Collector main
     window.

  o  Click on the specific query related to the
     information you are intersted in collecting.

  o  Click on Collect, in the middle of the Collector
     main window.

  o  Invoke the Analyzer to see the results of your
     performance data collection.

  Note that you can collect information from only one
  query at a time. That is, you must do a separate
  collection run for each query you choose. You can
  append the results of multiple collection runs to one
  performance data file, allowing you to analyze all your
  performance data together.

  For more information on collecting and displaying
  performance information, refer to:

  o  The online help, which is context-sensitive. This
     means you can click on any object in a PCA window
     and get help on it by simultaneously pressing
     the Help key on your keyboard while you press and
     hold MB1 on the mouse.For more information on using
     online help, see the Help menu on any PCA screen.

  o  The hardcopy manual, Guide to DIGITAL Performance
     and Coverage Analyzer for OpenVMS Systems

13  –  CCTPCAInfo frame

Close Help