SPOT man page


User Commands                                             spot(1)



NAME
     spot - run a tool chain on an  executable,  and  generate  a
     website for browsing the data


SYNOPSIS
     spot [ options ] [ target target-args | -p  pid ]


DESCRIPTION
     The spot command runs a set of performance tools on the tar-
     get  application  and  renders the output as a set of hyper-
     linked webpages. There are two ways that spot can be used:


     spot -P pid
          spot attaches to a running  process  and  gathers  data
          from the running process using a variety of probes.


     spot app params
          The application is run multiple times, each time  under
          a different probe.  So it is necessary for the applica-
          tion to be able to  run  multiple  times  without  user
          intervention.


     The webpages that spot produces record the following  infor-
     mation (if it is available).


     System information
          Information about the system which the  experiment  was
          collected on


     Build information
          Information about how the application was built


     Performance counter data
          Performance counter events recorded by the  CPU  during
          the  run  of  the application. This data can be used to
          indicate the types of events which are causing signifi-
          cant stall time.


     Instruction frequency data
          This indicates the mix of instructions executed by  the
          application (SPARC only).


     Time spent in code
          This data is from the collect command which returns the
          amount of time spent in the various routines. This data
          is rendered using er_html.


     Stall time profiles
          The collect command is used to profile the  application
          based on the performance counter events that contribute
          the largest stall time. The resulting profiles indicate
          where  in  the  program the stall events are occurring.
          This data is rendered using er_html.


     Execution counts
          This data indicates how often each routine  is  called,
          and  also  how often each individual instruction in the
          routine is called (SPARC only).


     System-wide bandwidth utilization
          Data  on  utilization  of  system-wide  utilization  of
          bandwidth for all running processes. It is not possible
          to attribute bandwidth to a particular  process  unless
          that  process  is the only one active at the time. This
          data  can  only  be  collected  if  the  user  has  the
          appropriate   permissions  to  access  the  performance
          counters on a system-wide basis. (SPARC only).


     Trap information
          Data on the number of traps encountered over the run of
          the  program. This information is only available if the
          user has the appropriate permissions to access trapstat
          data (SPARC only).


     If gnuplot is available on the path, then additional  graphs
     will be available that show the various events over time.


OPTIONS
     If invoked with no arguments, print  a  usage  message.   If
     /bin/perl is not installed on the system, fail with the mes-
     sage: "spot: Command not found", even though it is perl, not
     spot  that is missing.  If the version of perl on the system
     is not recent enough,  spot may fail with a message about an
     undefined variable.


     -c  path
          Specify a path for the  SunStudio  components  used  by
          spot.   If  both  SunStudio  and  spot are installed in
          their default locations, spot will find the  components
          it needs.  However, if either of them is installed in a
          different location, then this option  can  be  used  to
          specify  the  path.   This option is also useful if the
          user wants to override the default compiler and  use  a
          compiler  installed  in  a  location different from the
          default one.


     -D n Set the level of debug information to be  printed.  The
          default is 1. A setting of 0 means no output from spot,
          a setting of 2 means full debug information. The  debug
          output  will be available in debug.log in the completed
          report.


     -d directory
          Place output from spot in a subdirectory of the  speci-
          fied  directory.   If  not specified, the default is to
          place the data in the current directory.


     -h   Print help information.


     -o name
          Place the output from spot in  the  named  subdirectory
          named name<n> where n is the first name not in use.  If
          the name is specified, it may have any form, as long as
          the  directory  does  not  exist  at  the  time spot is
          invoked. If the name is not specified  then  spot  will
          default to using spot_run<n>


     -P pid
          Attach spot to a running process and produce report.


     -q   Suppress all spot output. (Equivanlent to -D 0)


     -T seconds
          This option is only valid when spot is attaching  to  a
          running  process.   It  sets  the number of seconds for
          which each of the probes should be attached to the pro-
          cess. The default is 300 seconds (5 minutes).


     -V   Print the current  version.   Do  not  examine  further
          arguments and perform no further processing.


     -v   Print the current version and further  detailed  debug-
          ging   information  about  the  conversion  being  run.
          (Equivalent to -D 2)


     -X   This option will cause spot to try to collect  extended
          information  about  the performance of the application.
          It will collect hardware counter profiles of the appli-
          cation  using those performance counters that have been
          identified by ripc as large contributors to the overall
          stall  time.  If possible, bandwidth and trap data will
          also be collected; the user needs the appropriate  per-
          missions for bandwidth and trap data to be collected.


SEE ALSO
     analyzer(1),    collect(1),     er_archive(1),     er_cp(1),
     er_export(1),  er_html(1),  er_mv(1), er_print(1), er_rm(1),
     er_src(1), and the Performance Analyzer manual.



                    Last change: January 2007                   4

Enter labels to add to this page:
Please wait 
Looking for a label? Just start typing.

Sign up or Log in to add a comment or watch this page.


The individuals who post here are part of the extended Sun Microsystems community and they might not be employed or in any way formally affiliated with Sun Microsystems. The opinions expressed here are their own, are not necessarily reviewed in advance by anyone but the individual authors, and neither Sun nor any other party necessarily agrees with them.

Copyright 1994-2009 Sun Microsystems, Inc.
Powered by Atlassian Confluence
Sun Guidelines on Public Discourse Privacy Policy Terms of Use Trademarks Site Map Employment Investor Relations Contact