Cube 4.x series

Download Build requirements Documentation ChangeLog

 

Cube, which is used as performance report explorer for Scalasca and Score-P, is a generic tool for displaying a multi-dimensional performance space consisting of the dimensions (i) performance metric, (ii) call path, and (iii) system resource. Each dimension can be represented as a tree, where non-leaf nodes of the tree can be collapsed or expanded to achieve the desired level of granularity. In addition, Cube can display multi-dimensional Cartesian process topologies.

The Cube 4.x series report explorer and the associated Cube4 data format is provided for Cube files produced with the Score-P performance instrumentation and measurement infrastructure or the Scalasca version 2.x trace analyzer (and other compatible tools). However, for backwards compatibility, Cube 4.x can also read and display Cube 3.x data.

This software is free but copyrighted
© 1998-2018 Forschungszentrum Jülich GmbH, Germany
© 2009-2015 German Research School for Simulation Sciences GmbH, Jülich/Aachen, Germany
By downloading and using this software you automatically agree to comply with the regulations as described in the Cube v4 license agreement.

Cube Packages

ATTENTION: Since version 4.4 Cube has been split in its components and is distributed in form of four separate packages: CubeW, CubeLib, CubeGUI and jCubeR.

Version Date Description
CubeW 4.4.1
10-Sep-2018
High performance C Writer library
Latest Release
MD5sum: 260cbcc72ea93378b5c2ea174b0fe87e
DOI

Includes:
  • Cleaned up public symbols in library.
  • Minor fixes.
CubeLib 4.4.2 10-Sep-2018
General purpose C++ library and tools
Latest Release
MD5sum: bda63f16ecbbcbe2a1383be79963309f
DOI
Includes:
    • Fixed bug in declustering algorithm leading sometimes to the wrong results.
    • Improved generation of the temporary directory.
    • cube_dump tool prints out the call path parameters.
    • cube_remap2 tool creates topology "Process x Thread [ x GPU ] and adds it to the result file.
    • Improvements of the tools guide:
      • elaborate purpose and details of missing tools;
      • describe details of the remapping tool;
      • describe syntax of .spec file.
    • ... for more details see ChangeLog
CubeGUI 4.4.2 10-Sep-2018
Graphical explorer
Latest Release
MD5sum: dc8075d0bf4d651aec38c4580294c348
DOI

Includes:
  • Improved "Find" functionality.
  • Improved automatic region display in the "Source Code viewer".
  • Fixed serious performance problem if remote X is used.
  • Improved GUI performance for large call trees. Show the number of call tree references in flat tree info.
    Switching between flat and call tree may cause the selection of a large amount of items in the call tree. This will be extremly slow, if more than 10.000 items are selected. The number of expanded and selected items is now limited, the items are marked instead and can be selected afterwards with context menu item "select marked"
  • 'Edit metric' is not disabled anymore, if creation of derived metric was cancelled.
  • Copy selected flat tree values as they are shown in the tree.
  • Renamed configure option --with-debug to --enable-debug to be more consistent with its semantic.
  • ... for more details see ChangeLog

Requires : Installed CubeLib 4.4.x
jCubeR 4.4 07-May-2018
Java reader library
Latest Release
MD5sum: d3fc35f04526b6a06333179aa2f3b151
DOI

Includes:
    • Build system improvements.
    • Documentation infrastructure refactoring.
Supplementary packages for download (Comfort zone)
Bundle 2.0
MD5sum : a57f582e9267a6a098a4a5ce26e515c3

NOTICE: Version of CubeBundle is separated from Cube components.
Version 2.0 contains CubeLib 4.4.2, CubeGUI 4.4.2 and CubeW 4.4.1
win32 binary
MD5sum : 7fae49f5942a623cac30f529067c608f
Mac OS .dmg
MD5sum : 60c6ce15a6beb5b6d7476c9462932834
Cube Java Reader (JAR)
MD5sum : a88881d37e67486aaa860f92b0947504
Latest trunk (Developers zone, NO WARRANTY )
CubeW
CubeLib
CubeGUI
jCubeR

 

Additional Cube Plugins

With 4.4 version of Cube we provide a set of additional plugins, which can be build and installed additionally to Cube GUI/Library 4.4. Note that they have additional prerequisites, which need to be installed before.

Please note that it is an EXPERIMENTAL development and we do not provide any warranty.
Cube Plugins were tested only on Linux Ubuntu with Qt 5.9.1 and have no documentation so far. Requirements (Score-P, OTF2, CubeGUI or similar) need to be available via PATH.

Plugin Name Date Description
ScorePion 0.2 10-Sep-2018
Score-P/Intel compiler filter creation plugin
Latest Release
MD5sum: 99e3bd3cecdf1b9a6d70897bd8ed9449


Allows to build/edit a filter file while exploring an experiment and selecting regions or files to be included or excluded in the filtering process. Allows to store the created filter file in Score-P format (.flt) or as an intel compiler filter file (.ifl)

Requires : Installed Score-P >=v4.0

Blade 07-May-2018
Simple OTF2 trace explorer
Latest Release
MD5sum: 305804fad578b1b6e4e188bd71b9b922


If experimental directory contains an OTF2 trace, this plugin allows to browse it and study communication patterns. Suitable only for small traces.

Requires : Installed Score-P >=v4.0 or OTF2 >=v2.1.1

JengaFett 0.2 10-Sep-2018
Metrics correlation explorer
Latest Release
MD5sum: 07a842a6a1fe9b7ca90457947f03d6fa


This plugin allows to display correlation between metrics as a parallel bar charts distribution across the system tree.

CallGraph 07-May-2018
Call graph display
Latest Release
MD5sum: eb48d4f7084cd5f6c53444eef0677fe4


This plugin builds a call graph based on call tree and assigns the metric values to the edges. Allows to detect a critical calls in the program execution.

Requires : Installed Graphviz

Supported Platforms

The Cube tool set has been successfully tested on the following platforms:

  • Linux (x86, x86_64, ia64, PPC, Power)
  • IBM BlueGene/Q
  • AIX (Power)
  • Cray
  • Mac OS X (x86_64) (10.8 and younger)
  • Windows 7 ( experimental )

In addition, the provided configure options (see installation instructions for Writer, Library and GUI) may provide a good basis for building and testing the toolset on other systems. Please report success/failure on other platforms to the Scalasca development team.