From any location, type:
> PROJECTIONS_LOCATION/bin/projections [NAME.sts]
where PROJECTIONS_LOCATION is the path to the main projections
directory.
Available options to the visualization component of PROJECTIONS include:
-h or -help: displays help information about available options.
-V or -version: displays current PROJECTIONS version number.
-u or -use-version : overrides the data interpretation
behavior of PROJECTIONS to explicitly use instead of the
current version. This is useful in scenarios where the latest version
of PROJECTIONS is not backward-compatible with older log formats.
-no-idle: tells PROJECTIONS to ignore idle time information
contained in the logs.
-bgsize : tells PROJECTIONS to compute
additional derived information by assuming a logical 3-D Torus
topology with the specified dimensionality and a processor-to-torus
placement policy that matches CHARM++'s placement policy on the BG/L
class of machines. The presence of this option enables additional
communication visualization options (see later). Note that this option
is really meant to be used for logs generated from virtual-node mode
BG/L executions. The additional information derived from any other
logs would probably be misleading.
-print_usage: tells PROJECTIONS to also write to standard
output the detailed graph numbers when viewing Usage Profiles (see
later). This is useful for generating visuals that are better
expressed by tools such as gnuPlot than through screen captures of
PROJECTIONS plots.
Supplying the optional NAME.sts file in the command line will
cause PROJECTIONS to load data from the file at startup. This shortcut
saves time selecting the desired dataset via the GUI's file dialog.
Figure 1:
PROJECTIONS main window
When PROJECTIONS is started, it will display a main window as shown
in figure 1. If summary (.sum) files are available in
the set of data, a low-resolution utilization graph (Summary Display)
will be displayed as shown. If summary files are not available, or if
PROJECTIONS was started without supplying the optional NAME.sts file, the main window will show a blank screen.
File contains 3 options. Open File(s) allows you to
explicitly load a data set. This happens if you had not specified a
NAME.sts file in the command line when starting PROJECTIONS
or if you wish to explicitly load a new dataset. It brings up a
dialog box that allows you to select the location of the dataset you
wish to study. Navigate to the directory containing your data and
select the .sts file. Click on `Open'. If you have selected a valid
file, PROJECTIONS will load in some preliminary data from the
files and then activate the rest of the options under the menu item
Tools. Close current data currently works the same way
as Close all data. They unload all current PROJECTIONS data
so one can load in a new set of data. They will also deactivate the
individual items provided in the Tools menu option.
Preferences generally allows you to set foreground or background
colors and entry method color schemes. This is useful for configuring
the color schemes of PROJECTIONS windows to be print-friendly.
Tools lists the set of available tools for analysis of generated
trace data. It will be described in great detail under section
4.2.2.
The Summary Display loaded on the Main Window displays basic processor
utilization data (averaged across all processors) over time
intervals. This is provided by the data generated by the summary
tracemode. This view offers no special features over and above the
Standard Graph Display described in section 4.4.
Please refer the appropriate section on information for using
its available features.
There should not be any serious performance issues involved in the
loading of summary data on the main window.
4.2.2 Available Tools
The following tools and views become available to you after a dataset
has been loaded (with the exception of Multirun Analysis) and may be
accessed via the menu item Tools:
The Graphs view is where you can analyze your data by breaking it
into any number of intervals and look at what goes on in each of those
intervals.
The Timelines view lets you look at what a specific processor is
doing at each moment of the program. It is the most detailed view of a
parallel application PROJECTIONS offers (and correspondingly, the
most resource-hungry).
The Usage Profile view lets you see percentage-wise what entry
methods each processor spends its time on during a specified time range.
It is particularly useful for identifying load imbalance and the probable
offending entry method.
The Communication view is a general tool that presents
communication properties contributed by each entry point across the
processors.
The Log File Viewer provides a human-readable, verbose
interpretation of a log file's entries.
The Histograms view presents entry point or communication
histogram information (ie. the frequency of occurrence of events given
an activity property like time bin size or message size on the
x-axis).
The Overview view gives user an overview of the utilization of
all processors during the execution. It is an extremely useful initial
tool to begin your performance analysis efforts with as it provides an
overall picture of application performance while being very
light-weight at the same time.
The Animation view animates the processor usage over a specified
range of time and a specified interval size.
The Time Profile Graph view is a more detailed presentation of
the Graphs utilization view in that it presents the time
contribution by each entry point across the desired time
interval. While the Graphs view can show the same data, it is
unable to stack the entry points, which proves useful in some cases.
