pyCPA is a pragmatic Python implementation of Compositional Performance Analysis (aka the SymTA/S approach provided by Symtavision (now: Luxoft)) used for research in worst-case timing analysis. Unlike the commercial SymTA/S tool, pyCPA is not intended for commercial-grade use and does not guarantee correctness of the implementation.
The following provides a brief overview on pyCPA. For more detailed Information, pleaser refer to the documentation.
Given, you have a (distributed) real-time system and you want to know about worst-case (end-to-end) timing behavior, then you can use pyCPA to obtain these bounds.
You provide your architecture in the form of resources such as busses and CPUs and the corresponding scheduling policies. In a second step, you define your task-graph which is a specification of task-communication (precedence relations) and tasks' properties (best/worst-case execution times, activation patterns).
pyCPA will then calculate the following metrics:
- worst-case response times (wcrt) of tasks
- end-to-end timing of task chains
- backlog of task activations (maximum buffer sizes)
- output event models of dependent tasks
- schedulers: static priority (non-)preemtive, round-robin, TDMA, FIFO
- task activation: periodic with jitter and minimum distance or generic events
- system analysis: event model propagation
- end to end analysis
- gantt-charts (spnp, spp only)
- graph of system topology
- SMFF support (through xml interface)
Why not? pyCPA is ideal for students who want to learn about real-time performance analysis research as well as researchers who want to extend existing algorithms. pyCPA is -as the name suggests- written in Python and extremely easy to use and extend. If you want, you can easily plugin new schedulers or your own analyses.
pyCPA should not be used in any commercial-grade, safety-critical designs. It does not provide analysis methods for commercial scheduling protocols like OSEK. Contact Luxoft if you need any commercial support for such applications.
- pyCPA cannot and will not obtain the worst-case execution time (WCET) of a task
- there is and will be no support for any specific protocols (e.g. OSEK, Ethernet, CAN, ARINC, AUTOSAR, etc.). Contact Luxoft if you need commercial support for any protocols.
Requirements:
- Python 2.7 or Python 3
- Python packages: setuptools, argparse, pygraphviz, matplotlib
- Optional python packages: numpy, simpy, xlrd
Download the latest version of the package, extract and install through setup.py.
Please refer to the more detailed installation instructions on readthedocs.org.
Run the examples and follow the documentation below.
The documentation is hosted on readthedocs.org.
You may also want to look at SMFF.
A list of related publications can be found at Bibliography. Please let us know about if you used/extended pyCPA for any of your publications so we can add it to the list.
More complex analysis and use cases are hosted in separate repositories, i.e.
- a task-chain analysis
- an analysis of the WATERS industrial challenge 2017