This package provides functionality which are common to front ends that translate application level programs into executables which run on a SpiNNaker machine.
In addition to a standard Python installation, this package depends on:
numpy
SpiNNUtilities
SpiNNMachine
PACMAN
SpiNNMan
spalloc
These dependencies can be installed using pip
:
pip install numpy
pip install SpiNNUtilities SpiNNMachine PACMAN SpiNNMan spalloc
If you want to use the spinnaker_router_provenance_mapper
command line tool
to visualise the traffic on SpiNNaker boards caused by your simulations, you
should install this package using:
pip install "SpiNNFrontEndCommon[plotting]"
This will additionally install matplotlib
and seaborn
to do the actual
plotting, as well as configuring the script.
To get plots of the traffic in your simulation, use the
spinnaker_router_provenance_mapper
command line tool (installed as described
above), passing in the name of a provenance database (usually called
provenance.sqlite3
and created within the run's reporting folders) that
contains the raw data. This will produce a number of graphs as images in your
current directory; those have fixed file-names.
spinnaker_router_provenance_mapper my_code/.../provenance.sqlite3
An example of the sort of map that might be produced is:
The P2P traffic being mapped is mainly used for system boot and control. The white square is due to a chip on that SpiNNaker board being marked as deactivated.
SpiNNFrontEndCommon python documentation
SpiNNFrontEndCommon C documentation