Go to main content
Index
Numbers and Symbols
- <JVM-System> function
<JVM-System> Function
- <no Java callstack recorded> function
<no Java callstack recorded> Function
- <Scalars> data object descriptor
<Scalars> Data Object
- <Total> data object descriptor
<Total> Data Object
- <Total> function
- comparing times with execution statistics
Comparisons of Timing Metrics
- described
<Total> Function
- <Truncated-stack> function
<Truncated-stack> Function
- <Unknown> function
- callers and callees
<Unknown> Function
- mapping of PC to
<Unknown> Function
- --jdkhome analyzer command option
-j | --jdkhome jvm-path
- -xdebugformat, setting debug symbol information format
Compiling to Analyze Source Code
- .er.rc file
Performance Analyzer Configuration File
- .er.rc file
-{cc,scc,dcc} com-spec- commands
Setting Defaults in .er.rc Files
Setting Defaults in .er.rc Files
- locations
Setting Defaults in .er.rc Files
- @plt function
Function Calls Between Shared Objects
A
- addpath command
addpath
path-list
- address spaces, text and data regions
Process Image
- aliased functions
Aliased Functions
- aliased hardware counters
Hardware Counter Lists
- aliases for hardware counters
Hardware Counter Lists
- alternate entry points in Fortran functions
Fortran Alternate Entry Points
- alternate source context
source|src {
filename
|
function-name
} [
N]
- Analyzer See
Performance Analyzer
- analyzer command
- data collection options
analyzer Command Options
- font size (-f) option
-f | --fontsize size
- help (-h) option
-? | -h | --help
- JVM options (-J) option
-J jvm-option
- JVM path (-j) option
-j | --jdkhome jvm-path
- verbose (-v) option
-v | --verbose
- version (-V) option
-V | --version
- annotated disassembly code See
disassembly code, annotated
- annotated source code See
source code, annotated
- API, Collector
Program Control of Data Collection Using libcollector Library
- appending path to files
addpath
path-list
- archiving load objects in experiments
archive
mode Subcommand
Archive Load Objects in Experiment with the -A
option
- artificial functions, in OpenMP call stacks
Artificial Functions
- attaching the Collector to a running process
Collecting Data From a Running Process With dbx on Oracle Solaris
Platforms
- attributed metrics
- defined
Function-Level Metrics: Exclusive, Inclusive, Attributed, and Static
- effect of recursion on
How Recursion Affects Function-Level Metrics
- illustrated
Call Tree Illustrating Exclusive, Inclusive, and Attributed Metrics
- use of
Function-Level Metrics: Exclusive, Inclusive, Attributed, and Static
B
- body functions, compiler-generated
- displayed by the Performance Analyzer
Compiler-Generated Body Functions
Compiler-Generated Body Functions
- names
Compiler-Generated Body Functions
- branch target
Branch Target
C
- call stack
Heap View- defined
Call Stacks and Program Execution
- effect of tail-call optimization on
Tail-Call Optimization
- incomplete unwind
Incomplete Stack Unwinds
- mapping addresses to program structure
Mapping Addresses to Program Structure
- unwinding
Call Stacks and Program Execution
- call stack fragment
Callers-Callees View
- call stack navigation
Called-By/Calls Panel
- Call Tree view
Call Tree View- filter data from context menu
Call Tree View
- Called-By/Calls panel
Called-By/Calls Panel
- callers-callees metrics
- attributed, defined
Function-Level Metrics: Exclusive, Inclusive, Attributed, and Static
- displaying list of in er_print utility
cmetric_list
- printing for a single function in er_print utility
csingle
function-name
[N]
- printing in er_print utility
callers-callees
- Callers-Callees view
Callers-Callees View
- clock profiling
- accuracy of metrics
Comparisons of Timing Metrics
- collecting data in dbx
profile
option
- collecting data with the collect command
Clock Profiling with the -p
option
- comparison with gethrtime and gethrvtime
Comparisons of Timing Metrics
- data in profile packet
Clock Profiling
- defined
Clock Profiling Under Oracle Solaris
- distortion due to overheads
Accuracy of Timing Metrics
- interval See
profiling interval
- metrics
Clock Profiling
Clock Profiling Under Oracle Solaris
- clock profiling in Timeline
Timeline View
- cloned functions
Cloned Functions
Cloned Functions
Cloned Functions
- collect command
- -I option
Specify Count Data Instrumentation Directory with I
directory
- -N option
Specify Excluded Libraries with N
library-name
- -P option
Attach to a Process with P
process-id
- archiving (-A) option
Archive Load Objects in Experiment with the -A
option
- –C option
Experiment Format
- clock profiling (-p) option
Clock Profiling with the -p
option
- collecting data with
Collecting Data Using the collect Command
- data collection options
Data Collection Options
- data limit (-L) option
Limit the Experiment Size with -L
size
- data race detection (-r) option
Data Race and Deadlock Detection with r
option
- dry run (-n) option
Try Out a Command with -n
Add a Comment to the Experiment with -C
comment
- experiment control options
Experiment Control Options
- experiment directory (-d) option
Set Directory of Experiment with -d
directory-name
- experiment group (-g) option
Create Experiment in a Group with -g
group-name
- experiment name (-o) option
Save the Command Output to a File with -O
file
- follow descendant processes (-F) option
Follow Processes with the -F
option
- hardware counter profiling (-h) option
Hardware Counter Profiling with collect -h
- heap tracing (-H) option
Heap Tracing Data with -H
option
- –i option
I/O Tracing with i
option
- Java version (-j) option
Profile Java with -j
option
- listing the options of
Collecting Data Using the collect Command
- –M option
MPI Profiling with M
option
- miscellaneous options
Other Options
- MPI tracing (-m) option
MPI Tracing with m
option
- output options
Output Options
- pause and resume data recording (-y) option
Signal Pause and Resume State with -y signal [ ,r]
- periodic sampling (-S) option
Sample Data with -S
option
- record count data (-c) option
Count Data with c
option
- record sample point (-l) option
Specify a Signal for Sampling with -l
signal
- stop target after exec (-x) option
Stop Profiled Target to Allow dbx attach with -x
- synchronization wait tracing (-s) option
Synchronization Wait Tracing with s Option
- syntax
Collecting Data Using the collect Command
- time range for data collection (–t) option
Set a Time Range with t
duration
- verbose (-v) option
Display Verbose Output with -v
- version (-V) option
Display the collect Version with -V
- with ppgsz command
Using collect With ppgsz
- ‐o option
Set Experiment Name with -o
experiment-name
- Collector
- API, using in your program
Java Interface
Program Control of Data Collection Using libcollector Library
- attaching to a running process
Collecting Data From a Running Process With dbx on Oracle Solaris
Platforms
- defined
Data the Collector Collects
Performance Data
Collector Tool
- disabling in dbx
disable Subcommand
- enabling in dbx
enable Subcommand
- running in dbx
To Run the Collector From dbx
- running with the collect command
Collecting Data Using the collect Command
- collectorAPI.h
Java Interface- as part of C and C++ interface to collector
C and C++ Interface
- common subexpression elimination
Common Subexpression Elimination
- compare experiments
Comparing Experiments- setting comparison style
Setting Comparison Style
- compiler commentary
Source View- classes defined
scc
com-spec
- cloned functions
Cloned Functions
- common subexpression elimination
Common Subexpression Elimination
- description of
Compiler Commentary
- filtering in er_src utility
-{cc,scc,dcc} com-spec
- filtering types displayed
Compiler Commentary
- inlined functions
Inlining of Functions
- loop optimizations
Loop Optimizations
- parallelization
Parallelization
- selecting for annotated disassembly listing in er_print utility
dcc
com-spec
- selecting for annotated source and disassembly listing in er_print utility
cc
com-spec
- selecting for annotated source listing in er_print utility
scc
com-spec
- compiler optimization
- inlining
Inlining of Functions
- parallelization
Parallelization
- compiler-generated body functions
- displayed by the Performance Analyzer
Compiler-Generated Body Functions
Compiler-Generated Body Functions
- names
Compiler-Generated Body Functions
- compiling
- affects of optimization on program analysis
Optimization at Compile Time
- debug symbol information format
Compiling to Analyze Source Code
- for dataspace profiling
Compiling for Dataspace and Memoryspace Profiling
- for Lines view
Compiling to Analyze Source Code
- Java programming language
Compiling Java Programs
- linking for data collection
Static Linking
- source code for annotated Source and Disassembly
Compiling to Analyze Source Code
- config.xml file
Performance Analyzer Configuration File
- configuration settings
Performance Analyzer Configuration File
- connecting to remote host
Connecting to a Remote Host in Performance Analyzer
- copying an experiment
Copying Experiments With the er_cp Utility
- correlation, effect on metrics
Accuracy of Timing Metrics
- CPI metric
Metric Name Strings
- CPU utilization samples in Timeline
Timeline View
- CPUs
- listing in CPUs view
CPUs View
- listing selected, in er_print utility
cpu_list
- metrics per CPU
CPUs View
- selecting in er_print utility
cpu_select
cpu-spec
- CPUs view
CPUs View
- custom filters
Using Advanced Custom Filters
- Cycles per instruction
Metric Name Strings
D
- data collection
- controlling from your program
Program Control of Data Collection Using libcollector Library
- disabling from your program
C, C++, Fortran, and Java API Functions
- disabling in dbx
disable Subcommand
- dynamic memory allocation effects
Using Dynamically Allocated Memory
- enabling in dbx
enable Subcommand
- from MPI programs
Collecting Data From MPI Programs
- MPI program, using the collect command
Running the collect Command for MPI
- pausing for collect command
Signal Pause and Resume State with -y signal [ ,r]
- pausing from your program
C, C++, Fortran, and Java API Functions
- pausing in dbx
pause Subcommand
- preparing your program for
Preparing Your Program for Data Collection and Analysis
- program control of
Program Control of Data Collection Using libcollector Library
- rate of
Estimating Storage Requirements
- resuming for collect command
Signal Pause and Resume State with -y signal [ ,r]
- resuming from your program
C, C++, Fortran, and Java API Functions
- resuming in dbx
resume Subcommand
- segmentation faults
Using Dynamically Allocated Memory
- using dbx
To Run the Collector From dbx
- using the collect command
Collecting Data Using the collect Command
- data objects
- <Scalars> descriptors
<Scalars> Data Object
- <Total>descriptor
<Total> Data Object
- defined
Mapping Data Addresses to Program Data Objects
- descriptors
Data Object Descriptors
- in hardware counter overflow experiments
data_objects
- layout
DataLayout View
- scope
Mapping Data Addresses to Program Data Objects
- data presentation
- setting options
Configuration Settings
- data races
- detailed information
rdetail
race-id
- listing
races
- Data Size view
Data Size View
- data types
Data the Collector Collects- clock profiling
Clock Profiling Under Oracle Solaris
- hardware counter profiling
Hardware Counter Profiling Data
- heap tracing
Heap Tracing (Memory Allocation) Data
- MPI tracing
MPI Tracing Data
- synchronization wait tracing
Synchronization Wait Tracing Data
- data-derived metrics
- displaying list of in er_print utility
data_metric_list
- data_layout command
data_layout
- data_objects command
data_objects
- data_single command
data_single
name
[N]
- DataLayout view
DataLayout View
- DataObjects view
DataObjects View
- dataspace profiling
Dataspace Profiling and Memoryspace Profiling- data objects
Mapping Data Addresses to Program Data Objects
- dbx collector subcommands
- archive
archive
mode Subcommand
- dbxsample
dbxsample { on | off }
- disable
disable Subcommand
- enable
enable Subcommand
- enable_once (obsolete)
status Subcommand
- hwprofile
hwprofile
option
- limit
limit
value Subcommand
- pause
pause Subcommand
- profile
profile
option
- quit (obsolete)
status Subcommand
- resume
resume Subcommand
- sample record
sample record
name Subcommand
- sample
sample
option
- show
show Subcommand
- status
status Subcommand
- store filename (obsolete)
status Subcommand
- store
store
option Subcommand
- synctrace
heaptrace
option
synctrace
option
- tha
tha
option
- dbx
- running the Collector in
To Run the Collector From dbx
- ddetail command
ddetail
deadlock-id
- deadlocks
- detailed information
ddetail
deadlock-id
- listing
deadlocks
- Deadlocks view
Deadlocks View
- deadlocks command
deadlocks
- default metrics
Functions View
- default path
setpath
path-list
- defaults
- setting in a defaults file
Setting Defaults in .er.rc Files
- descendant experiments
- loading
Starting Performance Analyzer
- setting mode for reading, in er_print utility
en_desc { on | off | =regexp}
- descendant processes
- collecting data for all followed
Follow Processes with the -F
option
- collecting data for selected
Collecting Data From a Running Process With dbx on Oracle Solaris
Platforms
- experiment location
Where the Data Is Stored
- experiment names
Experiments for Descendant Processes
- followed by Collector
Limitations on Data Collection for Descendant Processes
- limitations on data collection for
Limitations on Data Collection for Descendant Processes
- disassembly code, annotated
- branch target
Branch Target
- cloned functions
Cloned Functions
- description
Annotated Disassembly Code
- for cloned functions
Cloned Functions
Cloned Functions
- hardware counter metric attribution
Attribution of Hardware Counter Overflows
- HotSpot-compiled instructions
Dynamically Compiled Functions
- inclusive metrics
Inclusive Metrics
- instruction issue dependencies
Instruction Issue Grouping
- interpreting
Interpreting Annotated Disassembly
- Java native methods
Java Native Functions
- location of executable
Moving Experiments
- metric formats
Metric Formats
- printing in er_print utility
disasm|dis {
filename
|
function-name
} [
N]
- setting preferences in er_print utility
dcc
com-spec
- setting the highlighting threshold in er_print utility
dthresh
value
- st and ld instructions
Annotations for Store and Load Instructions
- viewing with er_src utility
Viewing Source/Disassembly Without an Experiment
- Disassembly view
Disassembly View
- disk space, estimating for experiments
Estimating Storage Requirements
- DTrace
- described
Kernel Experiments
- setting up access to
Setting Up Your System for Kernel Profiling
- Dual-Source view
Dual-Source View
- Duration view
Duration View
- dynamically compiled functions
- Collector API for
Dynamic Functions and Modules
- definition
Dynamically Compiled Functions
Dynamically Compiled Functions
E
- entry points, alternate, in Fortran functions
Fortran Alternate Entry Points
- environment variables
- JAVA_PATH
Limitations on Java Profiling
- JDK_HOME
Limitations on Java Profiling
- LD_BIND_NOW
Shared Object Handling
- LD_LIBRARY_PATH
Collecting Tracing Data From a Running Program
- LD_PRELOAD
Collecting Tracing Data From a Running Program
- PATH
Limitations on Java Profiling
- SP_COLLECTOR_NO_OMP
Limitations on OpenMP Profiling
- SP_COLLECTOR_STACKBUFSZ
<Truncated-stack> Function
Limitations on Data Collection
- SP_COLLECTOR_USE_JAVA_OPTIONS
Limitations on Java Profiling
- VampirTrace
MPI Tracing Data
- VT_BUFFER_SIZE
Storing MPI Experiments
MPI Tracing Data
- VT_MAX_FLUSHES
MPI Tracing Data
- VT_STACKS
Storing MPI Experiments
MPI Tracing Data
- VT_UNIFY
Storing MPI Experiments
- VT_VERBOSE
Storing MPI Experiments
MPI Tracing Data
- er_archive utility
er_archive Utility
- er_cp utility
Copying Experiments With the er_cp Utility
- er_export utility
er_export Utility
- er_heap.so, preloading
Collecting Tracing Data From a Running Program
- er_html utility
er_html Utility
- er_kernel utility
Kernel Experiments
Clock Profiling for the Oracle Solaris Kernel
- analyzing kernel profile
Analyzing a Kernel Profile
- hardware counter overflow profiling
Profiling the Kernel for Hardware Counter Overflows
- kernel and user processes
Profiling Kernel and User Processes
- use with collect and user processes
Alternative Method for Profiling Kernel and Load Together
- er_label utility
Labeling Experiments- command syntax
er_label Command Syntax
- time specification
er_label Command Syntax
- usage examples
er_label Examples
- usage in scripts
Using er_label in Scripts
- er_mv utility
Moving Experiments With the er_mv Utility
- er_print commands
- addpath
addpath
path-list
- allocs
allocs
- appendtfile
appendfile
filename
- callers-callees
callers-callees
- cc
cc
com-spec
- cmetric_list
cmetric_list
- cpu_list
cpu_list
- cpu_select
cpu_select
cpu-spec
- csingle
csingle
function-name
[N]
- data_layout
data_layout
- data_metric_list
data_metric_list
- data_objects
data_objects
- data_single
data_single
name
[N]
- dcc
dcc
com-spec
- ddetail
ddetail
deadlock-id
- deadlocks
deadlocks
- describe
describe
- disasm
disasm|dis {
filename
|
function-name
} [
N]
- dmetrics
dmetrics
metric-spec
- dsort
dsort
metric-spec
- en_desc
en_desc { on | off | =regexp}
- exit
exit
- experiment_list
experiment_list
- filter examples
Example Filter Expressions
- filter syntax
Expression Grammar
- filtering
Commands That Control Filtering of Experiment Data
- filters
filters
filter-exp
- fsingle
fsingle
function-name
[N]
- fsummary
fsummary
- functions
functions
- header
header
exp-id
- help
help
- ifreq
ifreq
- indx_metric_list
indx_metric_list
- indxobj
indxobj
indxobj-type
- indxobj_define
indxobj_define
indxobj-type
index-exp
- indxobj_list
indxobj_list
- keywords for filter expressions
Listing Keywords for a Filter Expression
- leaks
leaks
- limit
limit
n
- lines
lines
- lsummary
lsummary
- lwp_list
lwp_list
- lwp_select
lwp_select
lwp-spec
- metric_list
metric_list
- metrics
metrics
metric-spec
- name
name { long | short } [ :{
shared-object-name
|
no-shared-object-name
} ]
- object_api
object_api
object1,object2,...
- object_hide
object_hide
object1,object2,...
- object_list
object_list
- object_select
object_select
object1,object2,...
- object_show
object_show
object1,object2,...
- objects
objects
- objects_default
objects_default
- outfile
outfile {filename|-|--}
- overview
overview
exp_id
- pathmap
pathmap
old-prefix
new-prefix
- pcs
pcs
- procstats
procstats
- psummary
psummary
- quit
quit
- races
races
- rdetail
rdetail
race-id
- sample_list
sample_list
- sample_select
sample_select
sample-spec
- scc
scc
com-spec
- script
script
filename
- setpath
setpath
path-list
- sort
sort
metric_spec
- source
source|src {
filename
|
function-name
} [
N]
- statistics
statistics
exp_id
- sthresh
dthresh
value
sthresh
value
- thread_list
thread_list
- thread_select
thread_select
thread-spec
- usage examples
er_print Command Examples
- version
version
version
- viewmode
viewmode
{ user| expert | machine }
- er_print utility
- command-line options
er_print Syntax
- commands See
er_print commands
- metric keywords
Metric Visibility Characters
- metric lists
Metric Lists
- purpose
About er_print
- syntax
er_print Syntax
- er_rm utility
Deleting Experiments With the er_rm Utility
- er_src utility
Viewing Source/Disassembly Without an Experiment
- er_sync.so, preloading
Collecting Tracing Data From a Running Program
- event density
Timeline View
- event states
Timeline View
- events
- displayed in Timeline view
Timeline View
- exclusive metrics
- defined
Function-Level Metrics: Exclusive, Inclusive, Attributed, and Static
- for PLT instructions
Function Calls Between Shared Objects
- how computed
Call Stacks and Program Execution
- illustrated
Call Tree Illustrating Exclusive, Inclusive, and Attributed Metrics
- use of
Function-Level Metrics: Exclusive, Inclusive, Attributed, and Static
- execution statistics
- comparison of times with the function
Comparisons of Timing Metrics
- printing in er_print utility
statistics
exp_id
- experiment directory
- default
Where the Data Is Stored
- specifying in dbx
store
option Subcommand
- specifying with collect command
Set Directory of Experiment with -d
directory-name
- experiment groups
Where the Data Is Stored- creating
Starting Performance Analyzer
Experiment Groups
- default name
Experiment Groups
- defined
Experiment Groups
- multiple
Starting Performance Analyzer
- name restrictions
Experiment Groups
- preview
Starting Performance Analyzer
- removing
Deleting Experiments With the er_rm Utility
- specifying name in dbx
store
option Subcommand
- specifying name with collect command
Create Experiment in a Group with -g
group-name
- Experiment IDs view
Experiment IDs View
- experiment names
Where the Data Is Stored- default
Experiment Names
- kernel profiling
Subexperiments
Experiments on the Kernel and User Processes
- MPI default
Storing MPI Experiments
Experiments for MPI Programs
- restrictions
Experiment Names
- specifying in dbx
store
option Subcommand
- experiments See Also
experiment directory- appending current path
addpath
path-list
- archiving load objects in
archive
mode Subcommand
Archive Load Objects in Experiment with the -A
option
- copying
Copying Experiments With the er_cp Utility
- data aggregation
Starting Performance Analyzer
- default name
Experiment Names
- defined
Where the Data Is Stored
- descendant, loading
Starting Performance Analyzer
- groups
Experiment Groups
- header information in er_print utility
header
exp-id
- labeling
Labeling Experiments
- limiting the size of
limit
value Subcommand
Limit the Experiment Size with -L
size
- listing in er_print utility
experiment_list
- location
Where the Data Is Stored
- moving
Moving Experiments With the er_mv Utility
Moving Experiments
- multiple
Starting Performance Analyzer
- naming
Experiment Names
- opening
Starting Performance Analyzer
- preview
Starting Performance Analyzer
- remapping path prefixes
pathmap
old-prefix
new-prefix
- removing
Deleting Experiments With the er_rm Utility
- setting mode for Java and OpenMP
viewmode
{ user| expert | machine }
- setting path to find files
setpath
path-list
- storage requirements, estimating
Estimating Storage Requirements
- terminating from your program
C, C++, Fortran, and Java API Functions
- where stored
store
option Subcommand
Set Directory of Experiment with -d
directory-name
- Experiments view
Experiments View
- experiments, descendant
- setting mode for reading, in er_print utility
en_desc { on | off | =regexp}
- explicit multithreading
Explicit Multithreading
- export settings to .er.rc
Performance Analyzer Configuration File
- expression grammar
Expression Grammar
F
- fast traps
Traps
- File menu
Menu Bar
- filter data from context menu
- Call Tree view
Call Tree View
- Callers-Callees view
Callers-Callees View
- Functions view
Functions View
- filter expression examples
Example Filter Expressions
- filtering and library visibility
Setting Library and Class Visibility
- filtering experiment data
Using Filters
Filtering Data
- er_print
Commands That Control Filtering of Experiment Data
- with custom filters
Using Advanced Custom Filters
- with labels
Using Labels for Filtering
- Filters panel
Filtering Data
- Flame Graph
Flame Graph View
- font size
- changing
-f | --fontsize size
- Fortran
- alternate entry points
Fortran Alternate Entry Points
- Collector API
Program Control of Data Collection Using libcollector Library
- subroutines
Load Objects and Functions
- frames, stack See
stack frames
- function calls
- between shared objects
Function Calls Between Shared Objects
- in single-threaded programs
Single-Threaded Execution and Function Calls
- recursive, metric assignment to
How Recursion Affects Function-Level Metrics
- function list
- compiler-generated body function
Compiler-Generated Body Functions
- printing in er_print utility
functions
- sort order, specifying in er_print utility
sort
metric_spec
- function names, choosing long or short form in er_print utility
name { long | short } [ :{
shared-object-name
|
no-shared-object-name
} ]
- function PCs, aggregation
PCs View
Lines View
- function-list metrics
- displaying list of in er_print utility
metric_list
- selecting default in .er.rc file
dmetrics
metric-spec
- selecting in er_print utility
metrics
metric-spec
- setting default sort order in .er.rc file
dsort
metric-spec
- functions
- <JVM-System>
<JVM-System> Function
- <no Java callstack recorded>
<no Java callstack recorded> Function
- <Total>
<Total> Function
- <Truncated-stack>
<Truncated-stack> Function
- <Unknown>
<Unknown> Function
- @plt
Function Calls Between Shared Objects
- address within a load object
Load Objects and Functions
- aliased
Aliased Functions
- alternate entry points (Fortran)
Fortran Alternate Entry Points
- cloned
Cloned Functions
Cloned Functions
- Collector API
Dynamic Functions and Modules
Program Control of Data Collection Using libcollector Library
- definition of
Load Objects and Functions
- dynamically compiled
Dynamically Compiled Functions
Dynamically Compiled Functions
Dynamic Functions and Modules
- global
Aliased Functions
- inlined
Inlined Functions
- MPI, traced
MPI Tracing Data
- non-unique, names of
Non-Unique Function Names
- outline
Outline Functions
Outline Functions
- static, in stripped shared libraries
Static Functions
Static Functions From Stripped Shared Libraries
- static, with duplicate names
Non-Unique Function Names
- system library, interposition by Collector
Using System Libraries
- variation in addresses of
Process Image
- wrapper
Non-Unique Function Names
- Functions view
Functions View
H
- hardware counter
- defined
Hardware Counter Profiling Data
- hardware counter profiling
- collecting data with dbx
hwprofile
option
- defined
Hardware Counter Profiling Data
- hardware counter attribute options
Hardware Counter Profiling with collect -h
- hardware counter library, libcpc.so
Limitations on Hardware Counter Profiling
- hardware counter list
- aliased counters
Hardware Counter Lists
- description of fields
Hardware Counter Lists
- obtaining with collect command
Collecting Data Using the collect Command
- obtaining with dbx collector command
hwprofile
option
- raw counters
Format of the Raw Hardware Counter List
- hardware counter metrics, displayed in DataObjects view
DataObjects View
- hardware counter overflow profiling
- data in profile packet
Hardware Counter Overflow Profiling
- kernel
Profiling the Kernel for Hardware Counter Overflows
- hardware counter overflow value
- consequences of too small or too large
Hardware Counter Overflow Profiling
- defined
Hardware Counter Profiling Data
- setting in dbx
hwprofile
option
- hardware counter profiling
- collecting data with collect command
Hardware Counter Profiling with collect -h
- default counters
Hardware Counter Profiling with collect -h
- hardware counters
- choosing with dbx collector command
hwprofile
option
- counter names
Hardware Counter Profiling with collect -h
- data objects and metrics
data_objects
- list described
Hardware Counter Lists
- obtaining a list of
hwprofile
option
Collecting Data Using the collect Command
- overflow value
Hardware Counter Profiling Data
- heap size in Timeline
Timeline View
- heap tracing
- collecting data in dbx
heaptrace
option
- collecting data with collect command
Heap Tracing Data with -H
option
- metrics
Heap Tracing (Memory Allocation) Data
- preloading er_heap.so
Collecting Tracing Data From a Running Program
- Heap view
Heap View
- Help menu
Menu Bar
- high metric values
- in annotated disassembly code
dthresh
value
- in annotated source code
sthresh
value
I
- I/O tracing
I/O View- collecting data with collect command
I/O Tracing with i
option
- I/O view
I/O View
- inclusive metrics
- defined
Function-Level Metrics: Exclusive, Inclusive, Attributed, and Static
- effect of recursion on
How Recursion Affects Function-Level Metrics
- for outlined functions
Inclusive Metrics
- for PLT instructions
Function Calls Between Shared Objects
- how computed
Call Stacks and Program Execution
- illustrated
Call Tree Illustrating Exclusive, Inclusive, and Attributed Metrics
- use of
Function-Level Metrics: Exclusive, Inclusive, Attributed, and Static
- index lines
Index Lines in the Source View- in Disassembly tab
Annotated Disassembly Code
- in Disassembly view
Disassembly View
- in er_print utility
source|src {
filename
|
function-name
} [
N]
source|src {
filename
|
function-name
} [
N]
- in Source tab
Annotated Disassembly Code
Index Lines in the Source View
- in Source view
Source View
- index lines, special
- compiler-generated body functions
Compiler-Generated Body Functions
- HotSpot-compiled instructions
Dynamically Compiled Functions
- instructions without line numbers
Dynamically Compiled Functions
- Java native methods
Java Native Functions
- outline functions
Outline Functions
- index objects
indxobj
indxobj-type- defining
indxobj_define
indxobj-type
index-exp
- listing
indxobj_list
- Index Objects views
Index Objects Views
- index-object metrics
- displaying list of in er_print utility
indx_metric_list
- indxobj command
indxobj
indxobj-type
- indxobj_define command
indxobj_define
indxobj-type
index-exp
- indxobj_list command
indxobj_list
- inlined functions
Inlined Functions
- input file
- terminating in er_print utility
exit
quit
- to er_print utility
script
filename
- Inst–Freq view
Inst-Freq View
- instruction frequency
- printing list in er_print utility
ifreq
- instruction issue
- delay
Instruction Issue Delay
- grouping, effect on annotated disassembly
Instruction Issue Grouping
- Instructions per cycle
Metric Name Strings
- intermediate files, use for annotated source listings
Intermediate Files
- interposition by Collector on system library functions
Using System Libraries
- interval, profiling See
profiling interval
- interval, sampling See
sampling interval
- IPC metric
Metric Name Strings
J
- Java
- dynamically compiled methods
Dynamically Compiled Functions
Dynamic Functions and Modules
- profiling limitations
Limitations on Java Profiling
- setting er_print display output for
viewmode
{ user| expert | machine }
- Java Virtual Machine path, analyzer command option
-j | --jdkhome jvm-path
- JAVA_PATH environment variable
Limitations on Java Profiling
- JDK_HOME environment variable
Limitations on Java Profiling
K
- kernel profiling
- analyzing a profile
Analyzing a Kernel Profile
- clock profiling
To Profile the Kernel with er_kernel
- hardware counter overflow
Profiling the Kernel for Hardware Counter Overflows
- setting up your system for
Setting Up Your System for Kernel Profiling
- types of data
Kernel Experiments
Clock Profiling for the Oracle Solaris Kernel
- with collect and user processes
Alternative Method for Profiling Kernel and Load Together
- with user processes
Profiling Kernel and User Processes
- keywords, metric, er_print utility
Metric Visibility Characters
L
- labeling an experiment
Labeling Experiments
- labeling experiments
- and filtering
Using Labels for Filtering
- LD_LIBRARY_PATH environment variable
Collecting Tracing Data From a Running Program
- LD_PRELOAD environment variable
Collecting Tracing Data From a Running Program
- leaf PC, defined
Call Stacks and Program Execution
- leaks, memory, definition
Memory Allocation (Heap Tracing) Metrics
- libcollector.h
- as part of Java programming language interface to collector
Java Interface
- libcollector.so shared library
- preloading
Collecting Tracing Data From a Running Program
- using in your program
Program Control of Data Collection Using libcollector Library
- libcpc.so, use of
Limitations on Hardware Counter Profiling
- libfcollector.h
Fortran Interface
- libraries
- collectorAPI.h
Java Interface
- interposition on
Using System Libraries
- libcollector.so
Collecting Tracing Data From a Running Program
Program Control of Data Collection Using libcollector Library
- libcpc.so
Limitations on Hardware Counter Profiling
Using System Libraries
- MPI
Using System Libraries
- stripped shared, and static functions
Static Functions
Static Functions From Stripped Shared Libraries
- system
Using System Libraries
- Library and Class Visibility dialog box
Setting Library and Class Visibility
- limitations
- descendant process data collection
Limitations on Data Collection for Descendant Processes
- experiment group names
Experiment Groups
- experiment name
Experiment Names
- Java profiling
Limitations on Java Profiling
- profiling interval value
Limitations on Clock Profiling
- limiting output in er_print utility
limit
n
- limiting the experiment size
limit
value Subcommand
Limit the Experiment Size with -L
size
- Lines view
Lines View
- Load Machine Model button
Views Settings
- load objects
- addresses of functions
Load Objects and Functions
- contents of
Load Objects and Functions
- defined
Process Image
- listing selected, in er_print utility
object_list
- printing list in er_print utility
objects
- selecting in er_print utility
object_select
object1,object2,...
- symbol tables
Load Objects and Functions
- writing layouts of
data_layout
- loop optimizations
Loop Optimizations
- LWPs
- creation by Solaris threads
Explicit Multithreading
- listing selected, in er_print utility
lwp_list
- selecting in er_print utility
lwp_select
lwp-spec
M
- machine model
Mapping Performance Data to Memory Objects
Views Settings
- mangled function names
Cloned Functions
Compiler-Generated Body Functions
- memory allocations
Heap Tracing (Memory Allocation) Data- and leaks
Heap View
- effects on data collection
Using Dynamically Allocated Memory
- memory leaks, definition
Memory Allocation (Heap Tracing) Metrics
- memory object
- defining
- in Performance Analyzer
Views Settings
- with er_print
mobj_define
mobj-type
index-exp
- MemoryObjects views
MemoryObjects Views
- memoryspace profiling
Dataspace Profiling and Memoryspace Profiling
- methods See
functions
- metrics
- attributed
Callers-Callees View- See
attributed metrics
- clock profiling
Clock Profiling
Clock Profiling Under Oracle Solaris
- default
Functions View
- defined
Performance Data
- effect of correlation
Accuracy of Timing Metrics
- exclusive See
exclusive metrics
- function-list See
function-list metrics
- hardware counter, attributing to instructions
Attribution of Hardware Counter Overflows
- heap tracing
Heap Tracing (Memory Allocation) Data
- inclusive See
inclusive metrics
- inclusive and exclusive
Callers-Callees View
Functions View
- interpreting for instructions
Interpreting Annotated Disassembly
- interpreting for source lines
Interpreting Source Line Metrics
- memory allocation
Heap Tracing (Memory Allocation) Data
- MPI tracing
MPI Tracing Data
- static See
static metrics
- synchronization wait tracing
Synchronization Wait Tracing Data
- threshold
Disassembly View
- threshold, setting
Source View
- time precision
Functions View
- timing
Clock Profiling Under Oracle Solaris
- Metrics menu
Menu Bar
- metrics per experiment
Experiment IDs View
- Metrics tab
Metrics Settings
- microstates
Selection Details Window- contribution to metrics
Clock Profiling
- switching
Traps
- moving an experiment
Moving Experiments With the er_mv Utility
Moving Experiments
- MPI Chart Controls
MPI Chart Controls
- MPI Chart view
MPI Chart View
- MPI programs
- collecting data from
Collecting Data From MPI Programs
- collecting data with collect command
Running the collect Command for MPI
- experiment names
Storing MPI Experiments
Experiments for MPI Programs
- MPI Timeline Controls
MPI Timeline Controls
- MPI Timeline view
MPI Timeline View
- MPI tracing
MPI Tracing- collecting data with collect command
MPI Tracing with m
option
- functions traced
MPI Tracing Data
- metrics
MPI Tracing Data
- preloading the Collector library
Collecting Tracing Data From a Running Program
- multithreaded applications
- attaching the Collector to
Collecting Data From a Running Process With dbx on Oracle Solaris
Platforms
- multithreading
- explicit
Explicit Multithreading
N
- naming an experiment
Experiment Names
- networked disks
Where the Data Is Stored
- NFS
Where the Data Is Stored
- non-unique function names
Non-Unique Function Names
O
- OMP_preg command
OMP_preg
- OMP_task command
OMP_task
- Open Experiment dialog box
Performance Analyzer Configuration File
- OpenMP
- index objects, printing information for
OMP_task
OMP_preg
- metrics
OpenMP Metrics
- overview of execution
Overview of OpenMP Software Execution
- profiling data, machine representation
Machine View Mode of OpenMP Profiling Data
- profiling limitations
Limitations on OpenMP Profiling
- setting er_print display output for
viewmode
{ user| expert | machine }
- User mode call stacks
User Mode Call Stacks
- user mode display of profile data
User View Mode of OpenMP Profile Data
- OpenMP Parallel Region view
OpenMP Parallel Region View
- OpenMP parallelization
Parallelization
- OpenMP Task view
OpenMP Task View
- optimizations
- common subexpression elimination
Common Subexpression Elimination
- program analysis affect of
Optimization at Compile Time
- tail-call
Tail-Call Optimization
- options, command-line, er_print utility
er_print Syntax
- outline functions
Outline Functions
Outline Functions
- output file
- closing and opening new, in er_print utility
appendfile
filename
- closing, in er_print utility
outfile {filename|-|--}
- overflow value, hardware counter See
hardware counter overflow value
- overview data, printing in er_print utility
overview
exp_id
P
- parallel execution
- directives
Parallelization
- path prefix mapping
Pathmaps Settings
- path to files
setpath
path-list
- PATH environment variable
Limitations on Java Profiling
- pathmap command
pathmap
old-prefix
new-prefix
- Pathmaps tab
Pathmaps Settings
- pausing data collection
- for collect command
Signal Pause and Resume State with -y signal [ ,r]
- from your program
C, C++, Fortran, and Java API Functions
- in dbx
pause Subcommand
- PCs
- defined
Call Stacks and Program Execution
- from PLT
Function Calls Between Shared Objects
- ordered list in er_print utility
pcs
- PCs view
PCs View
- Performance Analyzer
- Call Tree view
Call Tree View
- Callers-Callees view
Callers-Callees View
- command-line options
-j | --jdkhome jvm-path
- CPUs view
CPUs View
- Data Size view
Data Size View
- DataLayout view
DataLayout View
- DataObjects view
DataObjects View
- Deadlocks view
Deadlocks View
- defined
Performance Analyzer Tool
- definition
About Performance Analyzer
- Disassembly view
Disassembly View
- Dual-Source view
Dual-Source View
- Duration view
Duration View
- Experiment IDs view
Experiment IDs View
- Experiments view
Experiments View
- Flame Graph view
Flame Graph View
- Functions view
Functions View
- I/O view
I/O View
- Index Objects views
Index Objects Views
- Inst–Freq view
Inst-Freq View
- Library and Class Visibility dialog box
Setting Library and Class Visibility
- Lines view
Lines View
- MemoryObjects views
MemoryObjects Views
- menu bar
Menu Bar
- Metrics tab
Metrics Settings
- MPI Chart Controls
MPI Chart Controls
- MPI Chart view
MPI Chart View
- MPI Timeline Controls
MPI Timeline Controls
- OpenMP Parallel Region view
OpenMP Parallel Region View
- OpenMP Task view
OpenMP Task View
- Pathmaps settings
Pathmaps Settings
- PCs view
PCs View
- Processes view
Processes View
- Races view
Races View
- recording an experiment
Starting Performance Analyzer
- Samples view
Samples View
- Search Path settings
Search Path Settings
- Seconds view
Seconds View
- Selection Details window
Selection Details Window
- Source view
Source View
- Source/Disassembly tab
Source/Disassembly Settings
- Source/Disassembly view
Source/Disassembly View
- starting
Starting Performance Analyzer
- Statistics view
Statistics View
- Summary tab
PCs View
- Threads view
Threads View
- Timeline settings
Timeline Settings
- Timeline view
Timeline View
Selection Details Window
- toolbar
Tool Bar
- views to be shown
Configuration Settings
- Visual Call Tree view
Flame Graph View
- performance data, conversion into metrics
Performance Data
- performance metrics See
metrics
- PLT (Program Linkage Table)
Function Calls Between Shared Objects
- ppgsz command
Using collect With ppgsz
- preloading
- er_heap.so
Collecting Tracing Data From a Running Program
- er_sync.so
Collecting Tracing Data From a Running Program
- libcollector.so
Collecting Tracing Data From a Running Program
- printing accumulated statistics from er_print utility
procstats
- process address-space text and data regions
Process Image
- Process-wide resource-utilization samples
Sample Data
- processes
- metrics per process
Processes View
- Processes view
Processes View
- Profile Application
- Preview command
Profiling Applications From Performance Analyzer
- profile packet
- clock data
Clock Profiling
- hardware counter overflow data
Hardware Counter Overflow Profiling
- size of
Estimating Storage Requirements
- synchronization wait tracing data
Synchronization Wait Tracing
- profiling interval
- defined
Clock Profiling Under Oracle Solaris
- experiment size, effect on
Estimating Storage Requirements
- limitations on value
Limitations on Clock Profiling
- setting with dbx collector command
profile
option
- setting with the collect command
profile
option
Clock Profiling with the -p
option
- profiling servers
Using er_label in Scripts
- profiling, defined
Data the Collector Collects
- program counter (PC), defined
Call Stacks and Program Execution
- program counter metrics
PCs View
- program execution
- call stacks described
Call Stacks and Program Execution
- explicit multithreading
Explicit Multithreading
- shared objects and function calls
Function Calls Between Shared Objects
- signal handling
Signals
- single-threaded
Single-Threaded Execution and Function Calls
- tail-call optimization
Tail-Call Optimization
- traps
Traps
- Program Linkage Table (PLT)
Function Calls Between Shared Objects
- program structure, mapping call stack addresses to
Mapping Addresses to Program Structure
R
- Races view
Races View
- races command
races
- raw hardware counters
Format of the Raw Hardware Counter List
Hardware Counter Lists
- rdetail command
rdetail
race-id
- recursive function calls
- metric assignment to
How Recursion Affects Function-Level Metrics
- remapping path prefixes
pathmap
old-prefix
new-prefix
Pathmaps Settings
- remote Performance Analyzer
Using Performance Analyzer Remotely- client requirements
Using Performance Analyzer on a Desktop Client
- connecting to remote host
Connecting to a Remote Host in Performance Analyzer
- server requirements
Using Performance Analyzer on a Desktop Client
- removing an experiment or experiment group
Deleting Experiments With the er_rm Utility
- restrictions See
limitations
- resuming data collection
- for collect command
Signal Pause and Resume State with -y signal [ ,r]
- from your program
C, C++, Fortran, and Java API Functions
- in dbx
resume Subcommand
S
- sample code
About Performance Analyzer
Performance Analyzer Tool
- samples
- circumstances of recording
Sample Data
- defined
Sample Data
- displayed in Timeline view
Timeline View
- information contained in packet
Sample Data
- interval See
sampling interval
- listing selected, in er_print utility
sample_list
- manual recording in dbx
sample record
name Subcommand
- manual recording with collect
Specify a Signal for Sampling with -l
signal
- metrics
Samples View
- periodic recording in dbx
sample
option
- periodic recording with collect command
Sample Data with -S
option
- recording from your program
C, C++, Fortran, and Java API Functions
- recording when dbx stops a process
dbxsample { on | off }
- selecting in er_print utility
sample_select
sample-spec
- Samples view
Samples View
- sampling interval
- defined
Sample Data
- setting in dbx
sample
option
- Search Path settings
Search Path Settings
- seconds
- metrics recorded per second
Seconds View
- Seconds view
Seconds View
- segmentation faults during data collection
Using Dynamically Allocated Memory
- Selection Details window
Selection Details Window
- setpath command
setpath
path-list
- setting mode for reading descendant experiments in er_print utility
en_desc { on | off | =regexp}
- settings
Performance Analyzer Configuration File
- setuid, use of
Using setuid and setgid
- shared objects, function calls between
Function Calls Between Shared Objects
- Show/Hide/APIonly Functions dialog box, see Library and Class Visibility dialog box
Setting Library and Class Visibility
- signal handlers
- installed by Collector
Signals
Data Collection and Signals
- user program
Data Collection and Signals
- signals
- calls to handlers
Signals
- profiling
Data Collection and Signals
- profiling, passing from dbx to collect command
Stop Profiled Target to Allow dbx attach with -x
- use for manual sampling with collect command
Specify a Signal for Sampling with -l
signal
- use for pause and resume with collect command
Signal Pause and Resume State with -y signal [ ,r]
- single-threaded program execution
Single-Threaded Execution and Function Calls
- sort order
- function list, specifying in er_print utility
sort
metric_spec
- source and disassembly code, annotated
- setting preferences in er_print utility
cc
com-spec
- source and object files used in experiments
Search Path Settings
- source code, annotated
- cloned functions
Cloned Functions
- compiler commentary
Compiler Commentary
- compiler-generated body functions
Compiler-Generated Body Functions
- description
Source Line Metrics
Annotated Source Code
- discerning annotations from source
Identifying the Original Source Lines
- for cloned functions
Cloned Functions
- index lines
Index Lines in the Source View
- instructions without line numbers
Dynamically Compiled Functions
- interpreting
Interpreting Source Line Metrics
- location of source files
Moving Experiments
- metric formats
Metric Formats
- outline functions
Outline Functions
- printing in er_print utility
source|src {
filename
|
function-name
} [
N]
- setting compiler commentary classes in er_print utility
scc
com-spec
- setting the highlighting threshold in er_print utility
sthresh
value
- use of intermediate files
Intermediate Files
- viewing in Performance Analyzer
Annotated Source Code
- viewing with er_src utility
Viewing Source/Disassembly Without an Experiment
- source code, compiler commentary
Source View
- source lines, ordered list in er_print utility
lines
- Source view
Source View
- Source/Disassembly tab
- in Settings
Source/Disassembly Settings
- Source/Disassembly view
Source/Disassembly View
- SP_COLLECTOR_STACKBUFSZ environment variable
<Truncated-stack> Function
Limitations on Data Collection
- stack depth
Limitations on Data Collection
- stack frames
- defined
Single-Threaded Execution and Function Calls
- from trap handler
Traps
- reuse of in tail-call optimization
Tail-Call Optimization
- static functions
- duplicate names
Non-Unique Function Names
- in stripped shared libraries
Static Functions
Static Functions From Stripped Shared Libraries
- static linking
- effect on data collection
Static Linking
- static metrics
Static Metric Information- defined
Function-Level Metrics: Exclusive, Inclusive, Attributed, and Static
- Statistics view
Statistics View
- storage requirements, estimating for experiments
Estimating Storage Requirements
- subroutines See
functions
- summary metrics
- for a single function, printing in er_print utility
fsingle
function-name
[N]
- for all functions, printing in er_print utility
fsummary
- Summary tab
PCs View
- symbol tables, load-object
Load Objects and Functions
- synchronization delay event count
- metric defined
Synchronization Wait Tracing Metrics
- synchronization delay events
- data in profile packet
Synchronization Wait Tracing
- defined
Synchronization Wait Tracing Data
- synchronization wait time
- defined
Synchronization Wait Tracing
Synchronization Wait Tracing Data
- metric, defined
Synchronization Wait Tracing Metrics
- synchronization wait tracing
- collecting data in dbx
synctrace
option
- collecting data with collect command
Synchronization Wait Tracing with s Option
- data in profile packet
Synchronization Wait Tracing
- defined
Synchronization Wait Tracing Data
- metrics
Synchronization Wait Tracing Data
- preloading er_sync.so
Collecting Tracing Data From a Running Program
- threshold See
threshold, synchronization wait tracing
- wait time
Synchronization Wait Tracing
Synchronization Wait Tracing Data
- syntax
- er_archive utility
er_archive Utility
- er_export utility
er_export Utility
- er_html utility
er_html Utility
- er_print utility
er_print Syntax
- er_src utility
Viewing Source/Disassembly Without an Experiment
- system-wide profiling
Profiling Kernel and User Processes
T
- tail-call optimization
Tail-Call Optimization
Tail-Call Optimization
- threads
- creation of
Explicit Multithreading
- listing selected, in er_print utility
thread_list
- metrics
Threads View
- selecting in er_print utility
thread_select
thread-spec
- worker
Explicit Multithreading
- Threads view
Threads View
- threshold, highlighting
- in annotated disassembly code, er_print utility
dthresh
value
- in annotated source code, er_print utility
sthresh
value
- threshold, synchronization wait tracing
- calibration
Synchronization Wait Tracing Data
- defined
Synchronization Wait Tracing Data
- effect on collection overhead
Synchronization Wait Tracing
- setting with dbx collector
synctrace
option
- setting with the collect command
synctrace
option
- time metrics, precision
Functions View
- Timeline settings
Timeline Settings
- Timeline view
Timeline View- Selection Details window
Selection Details Window
- TLB (translation lookaside buffer) misses
Attribution of Hardware Counter Overflows
Traps
- toolbar
Tool Bar
- Tools menu
Menu Bar
- tracing data in Timeline
Timeline View
- traps
Traps
- tutorials
About Performance Analyzer
Performance Analyzer Tool
U
- unwinding the call stack
Call Stacks and Program Execution
- User mode
Formats Settings
- User mode call stacks in OpenMP applications
User Mode Call Stacks
V
- VampirTrace
MPI Tracing Data
- version information
- for collect command
Display the collect Version with -V
- for er_cp utility
Copying Experiments With the er_cp Utility
- for er_mv utility
Moving Experiments With the er_mv Utility
- for er_print utility
version
- for er_rm utility
Deleting Experiments With the er_rm Utility
- for er_src utility
-V
- view mode
- explained
Formats Settings
- viewmode command
viewmode
{ user| expert | machine }
- views
- selecting for display
Views Settings
- Views menu
Menu Bar
- Views settings
Views Settings
- Visual Call Tree
Flame Graph View
W
- wait time See
synchronization wait time
- wrapper functions
Non-Unique Function Names