====== Vampir ======
---- dataentry analysisTool ----
name           : Vampir # 
features_ttags : performance analysis, application usage # 
layers_ttags   : application, operating system, MPI # 
version        : 9.0 (2015-11-30) # 
webpage_url    : https://www.vampir.eu/ # 
license_ttag   : commercial # 
----


VampirTrace is a tool for inspecting Trace-files (in Open Trace Format) of parallel applications. It allows users to perform analysis on performance problems and understand bottlenecks in their parallel programs.
In the context of parallel-I/O Vampir can help finding out if and where I/O is actually a bottleneck in your program.

====== Usage ======
Compile your program with the Vampir Compiler to enable writing of trace files:
<code>vtcc you_file.c</code>

The program is executed in the usual way. Running it produces a trace-file that can be opened in the Vampir gui.

As pictured below Vampir displays a timeline and a 'Function Summary' showing which parts of the program take the longest.
Using the function summary it is possible to see whether I/O actually takes up a large part of the programs running time. The timeline allow the user to inspect whether processes wait for I/O operations to finish.


vtcc is only a wrapper around a compiler, this allows us to switch out the underlying compiler when needed.
This makes it possible to compile mpi-applications with tracing enabled.
<code>vtcc -vt:cc mpicc</code>

The Vampir GUI offers a variety of other functionality to analyze your programs I/O behavior.
Check out the [[https://www.vampir.eu/tutorial/manual|official documentation]]!

====== Output ======
{{ :tools:analysis:vampir-io.png |}}


===== Similar tools =====

---- datarelated ----
cols  : features, layers
max   : 5
sort  : ^name
----