Panda

Panda

Panda transforms the PRDG description of an algorithm into a network of parallel running Processes, the desired Kahn Process Network. This tool uses the Polylib library extensively. Using Polylib, Panda can generate Ehrhart polynomials that give a symbolic expression for the number of integer points available in arbitrary polyhedra. The conversion from the PRDG to a Process Network happens in three steps: domain scanning, domain reconstruction and linearization.

Panda can generate two different Process Network representations: static and dynamic.

Static:
In the static representation, only information is given about how many times a particular process in a Process Network will fire and how many tokens will be communicated with other actors. This information is given as Ehrhart polynomials, which can be evaluated in terms of the original parameters of the Matlab description. This allows for very fast simulation of many Process Networks, for different parameters values,

Dynamic:
In the dynamic representation, the Fire method of an actor is generated. In this code-generation process, the polyhedra in the PRDG are converted into for-next loops and if/else statements. The dynamic description can be used to perform a functional simulation of a Process Network.

A couple of examples are given below of the results Panda produces. It includes the XML-file of the process network description, the description of the actors and screen dump of the result. The results are visualized using the Vergil visual editor of Ptolemy II.

Example 1, QR

Example 2, Faddeev

Example 3, the SVD algorithm cannot be shown.

The current implementation of Polylib still has difficulties to handle lattices, leading to degenerated domain descriptions that causes Polylib to crash. Vincent Loechner is working on fixing this problem.