Visible to Intel only — GUID: GUID-3A526C17-B421-45A7-9A5C-1D730D0A81DF
Tracing Conventional MPI Applications
Tracing Failing MPI Applications
Tracing OpenSHMEM* Applications
Tracing MPI File IO
Handling of Communicator Names
Tracing MPI Load Imbalance
Tracing User Defined Events
Configuring the Collector
Filtering Trace Data
Recording OpenMP* Regions Information
Tracing System Calls (Linux* OS)
Collecting Lightweight Statistics
Recording Source Location Information
Recording Hardware Performance Information (Linux* OS)
Recording Operating System Counters
Tracing Library Calls
Examples
Correctness Checking
Tracing Distributed Non-MPI Applications
ACTIVITY
ALTSTACK
AUTOFLUSH
CHECK
CHECK-LEAK-REPORT-SIZE
CHECK-MAX-DATATYPES
CHECK-MAX-ERRORS
CHECK-MAX-PENDING
CHECK-MAX-REPORTS
CHECK-MAX-REQUESTS
CHECK-SUPPRESSION-LIMIT
CHECK-TIMEOUT
CHECK-TRACING
CLUSTER
COMPRESS-RAW-DATA
COUNTER
CURRENT-DIR
DEADLOCK-TIMEOUT
DEADLOCK-WARNING
DEMANGLE
DETAILED-STATES
ENTER-USERCODE
ENVIRONMENT
EXTENDED-VTF
FLUSH-PID
FLUSH-PREFIX
GROUP
HANDLE-SIGNALS
INTERNAL-MPI
KEEP-RAW-EVENTS
LOGFILE-FORMAT
LOGFILE-NAME
LOGFILE-PREFIX
LOGFILE-RANK
MEM-BLOCKSIZE
MEM-FLUSHBLOCKS
MEM-INFO
MEM-MAXBLOCKS
MEM-MINBLOCKS
MEM-OVERWRITE
NMCMD
OS-COUNTER-DELAY
PCTRACE
PCTRACE-CACHE
PCTRACE-FAST
PLUGIN
PROCESS
PROGNAME
PROTOFILE-NAME
STATISTICS
STATE
STF-PROCS-PER-FILE
STF-USE-HW-STRUCTURE
STOPFILE-NAME
SYMBOL
SYNC-MAX-DURATION
SYNC-MAX-MESSAGES
SYNC-PERIOD
SYNCED-CLUSTER
SYNCED-HOST
TIME-WINDOWS (Experimental)
TIMER
TIMER-SKIP
UNIFY-COUNTERS
UNIFY-GROUPS
UNIFY-SCLS
UNIFY-SYMBOLS
VERBOSE
VT_START_PAUSED
VT_COMPRESS_TRACE
Parameter Checking
Premature Exit
Overlapping Memory
Detecting Illegal Buffer Modifications
Buffer Given to MPI Cannot Be Read or Written
Distributed Memory Checking
Illegal Memory Access
Request Handling
Datatype Handling
Buffered Sends
Deadlocks
Checking Message Transmission
Datatype Mismatches
Data Modified during Transmission
Checking Collective Operations
Freeing Communicators
Process Aggregation
Function Aggregation
Function Group Color Editor
Filtering Dialog Box
Tagging Dialog Box
Idealization Dialog Box
Imbalance Diagram Dialog Box
Trace Merge Dialog Box
Details Dialog Box
Source View Dialog
Time Interval Selection
Configuration Dialogs
Find Dialog Box
Command line for Intel® VTune™ Profiler and Intel® Advisor Dialog Box
OTF2 to STF Conversion Dialog Box
Configuration Assistant
Visible to Intel only — GUID: GUID-3A526C17-B421-45A7-9A5C-1D730D0A81DF
Tracing Library Calls
If you have an application that makes heavy use of libraries or software components developed independently, you may want to exclude the information not related directly to your application from the trace data. At the same time, the library developer might want to do the opposite – trace only data related to their library.
Intel® Trace Collector provides a capability to turn off tracing for functions at a certain call stack level, that is to fold them. If you want to trace calls within the folded functions, you can unfold them.
To enable folding, use the FOLD and UNFOLD keywords for the STATE, SYMBOL or ACTIVITY configuration options to select functions for folding by their name (SYMBOL), class (ACTIVITY) or both (STATE). Use the CALLER keyword to specify the function caller. See Filtering Trace Data for details on syntax.
NOTE:
To enable Intel Trace Collector to profile non-MPI functions, make sure to instrument them using the compiler instrumentation or API. See Tracing User Defined Events.
Below are examples of folding for the application with four additional libraries.
General Structure of an Application Using Multiple Libraries
From the figure above, the following information may be of interest for the application and the library developers:
Application developer
lib1, lib2, lib4 are called by the application. The application developer codes these calls and can change the sequence and parameters to them to improve performance (arrows "1").
lib3 is never called directly by the application. The application developer has no way to tailor the use of lib3, therefore these calls (arrows "3") are of no interest to him.
lib4 is called both directly by the application, and indirectly through lib2. Only the direct use of lib4 can be influenced by the application developer, therefore is of interest to them.
Library developer
The lib2 developer will need information about the calls from the application, to component libraries (lib3 and lib4), and to system-level services (MPI). They will have no interest in performance data for lib1. The lib1 developer will have no interest in data from lib2, lib3, and lib4.
Examples
In this section folding is illustrated by giving configurations that apply to the example above. The sample libraries.c program reproduces the same pattern. Its call tree looks as follows (calls are aggregated and sorted by name, therefore the order is not sequential):
By using the configuration options listed below, different parties can run the same executable to get different traces:
Application developer: Trace the application only with the top-level calls in lib1, lib2, and lib4.
Configuration file: run_splibraries_app.conf
STATE lib*:* FOLD
Call tree:
lib2 developer: Trace only calls in lib2, including its top-level calls
Configuration file: run_splibraries_lib2.conf
Call tree:
lib2 Developer, detailed view: Trace the top-level calls to lib2 and all lib2, lib3, lib4 and system services invoked by them
Configuration file: run_splibraries_lib2detail.conf
STATE Application:* FOLD STATE lib2:* UNFOLD
Call tree:
Application and lib4 Developers: Trace the calls in lib4 only made by the application
Configuration file: run_splibraries_lib4.conf
STATE *:* FOLD STATE lib4:* UNFOLD CALLER Application:*
Call tree:
It is assumed that the application, library and system calls are instrumented in the way that their classes are different. Alternatively, you can match against the function name prefix that is shared by all library calls in the same library.