Third Party Uses of the Ptolemy Software Environment

Current

Historical

Current

• Universidad de Vigo, Spain, created BerbeX, a COFDM-based digital TV transmission systems simulator, which uses Ptolemy.

• BNeD, Broadband Network Design, in Berlin, Germany, has developed and sells a library (BroadNeD) for the simulation of optical telecommunication networks. The optical devices are modeled from the systems point of view using advanced algorithmic modeling and allow in their totality the simulation of next generation optical networks with different network architectures. Ptolemy is used as the simulation environment. On February 29, 1998, BNeD and HP announced a plan to develop fiber optics software. (Remote copy of the announcement - Local copy)

• Cadence has developed a new architecture for "full-system, mixed-level, and mixed-domain simulation" based on research from the Ptolemy Project. They leveraged Ptolemy's "system-level design framework that allows mixing of multiple models of computation". Cadence describes their new CONVERGENCE Simulation Architecture in a press release. The computation model and scheduling algorithms in Cadence's Signal Processing WorkSystem were heavily influenced by Ptolemy's Synchronous Dataflow domain and its scheduling algorithms.

• Dresden University of Technology has develop WiNeS - Wireless Network System Simulator. The Ptolemy 0.7.1 Discrete Event Contrib palette includes a Tk demonstration of a WiNeS prototype. Contact: Jens Voigt

• HP EESof, a division of Hewlett Packard has a version of HP DSP Designer, which is based on research from the Ptolemy Project. The press release announcing the product summarizes its capabilities.

• Lyre sells a Motororola 56x/Xilinx 4000 standalone box that uses Ptolemy under Linux.

• Sanders, a Lockheed Martin Company is extending Ptolemy as part of a Algorithm Analysis and Mapping DARPA research project.

• SSS, Structured Software Systems, is working on automated securities trading systems based on Ptolemy. SSS uses a derivative of the DE domain for both historical testing and live execution of trading systems. SSS has also made numerous small improvements in the Ptolemy infrastructure, most of which have found their way back to Berkeley for inclusion in future releases. Contact: Frank Key or Tom Lane.

• Thompson, CSF, in Orsay, France, has developed their own Ptolemy domain and performed a valuable evaluation of Ptolemy. A summary their work with Ptolemy is available.
  • They have developed a multidimensional dataflow domain called ArrayOL which is based on the Ilog Views C++ graphical user-interface library. Contact: Xavier Warzee

  • They have performed an in-depth Evaluation of Ptolemy 0.4.1. Contact: Xavier Warzee

• The UC Berkeley CAD group is using Ptolemy as a framework for the POLIS system, which is a hardware/software codesign package, with application to automatic control. Contact: polis-questions@ic.eecs.berkeley.edu

• University of Pittsburgh Optical Computing Group and University of California at San Diego OptoElectronic Computing Group use Ptolemy in designing and simulating Free Space Optoelectronic Information Processing Systems.
  • Computer Aided Design and Simulation of Free Space Optoelectronic Information Processing Systems. Contact: Steven P. Levitan

• The University of Texas at Austin uses Ptolemy in both research and education.
  • Srikanth Gummadi is using Ptolemy to develop, simulate, and synthesize real-time constant modulus arrays for smart antennas. The arrays can be modelled completely in the Synchronous Dataflow model of computation. Contact: Srikanth Gummadi.

  • Ashutosh Kulkarni is using Ptolemy in benchmarking code generation methodologies. Contact: Ashutosh Kulkarni.

  • Biao Lu is using Ptolemy for modeling, simulating, and synthesizing heterogeneous sytems composed of neural network and signal/image processing subsystems by mixing dataflow models. Most (artificial) neural network and signal/image processing subsystems can be modeled using Synchronous Data. Some neural networks, such as Cellular Neural Networks, and some signal processing subsystems, such as timing recovery in modems, require Boolean Dataflow. Contact: Biao Lu.

  • Prof. Brian Evans uses Ptolemy as a laboratory for students in Embedded Software Systems to explore computational models, synthesis of embedded software, and heterogeneous system simulation. Contact: Prof. Brian Evans.

• White Eagle Systems Technology, Inc., uses Ptolemy in designing systems and developing CAD tools.
  • Hardware/software codesign of signal processing systems for speech, RADAR, networking and telecommunications. They have developed their own Ptolemy code generation domains for various DSP processors. Contact: Richard Tobias

  • Developed a dataflow CAD tool derived from research by the Ptolemy and Hyper projects at U.C. Berkeley. The tool assigned, scheduled, and generated code for the Mercury RACE architecture. The tool was used to develop radar and sonar applications on Mercury systems with over 128 Intel i860 processors. Contact: Richard Tobias

Historical

• BDTI, Berkeley Design Technology, Inc., in Fremont, CA, developed a prototype of a layer on top of the Ptolemy kernel called Ptolemy HSIM (Heterogeneous Simulation). Ptolemy HSIM served as a simulation backplane that allows Cadence's Signal Processing WorkSystem (SPW), Cadence's Bones, and Precedence's SimMatrix environments to cooperate during a simulation. SimMatrix is a synchronization mechanism for connecting 30 different VHDL and Verilog simulators together. The Ptolemy HSIM prototype was developed by BDTI for Lockheed-Martin, but there are no plans to release Ptolemy HSIM. For further information, contact Bernie Schaming at Lockheed-Martin.

• BU, Boston University, in Boston, MA, through its Knowledge-Based Signal Processing Group, has developed an environment called the Integrated Processing and Understanding of Signals (IPUS) that has been encapsulated into Ptolemy as a new domain. Ptolemy serves as the organizing framework and provides a computational engine through its other domains. IPUS can apply a family of knowledge-based techniques for iteratively refining a computation by dynamically selecting the algorithms to be applied to the data on the basis of results provided by previous algorithms. Thus, IPUS is a reasoning framework for signal reprocessing, incremental refinement, and signal understanding. Contact: Joe Winograd and Prof. S. Hamid Nawab

• DQDT, Dimensions in Quick Design Turnaround, in Carlsbad, CA, has derived a new VHDL domain to serve as a front-end specification and VHDL code generation environment for behavioral modeling and synthesis of Application-Specific Integrated Circuits. They applied the same approach using Mentor Graphics DSP Station as a front end. They have also customized the edit parameters dialogue to support Tcl/Tk widgets. For example, instead of typing in YES or NO for a boolean parameter, you could click on a yes or no button. They have also modified the schematic entry to support connections that are made by name (reference) instead of by creating a wire. Contact: Thomas M. Cesear or Scott R. Powell.

• FOKUS, Research Institute for Open Communication Systems, in Berlin, Germany, has developed a TCP Simulator using Ptolemy. Dorgham Sisalem's Master's Thesis describes the work.

• Sanders, Rapid-Prototyping Group, in Nashua, NH, has developed three different extensions to Ptolemy, as a prime contractor on the ARPA Rapid Prototyping of Application Specific Signal Processors (RASSP) Project.
  • They have written their own graphical front-end to Ptolemy that allows a user to sketch a target architecture and quickly map the stars in an SDF graph to the processors in the architecture. Extensions to the DE domain have been implemented to allow a performance-level model of the architecture to be simulated. They create a DE domain model representing the mapping of the algorithm to the architecture and use the Ptolemy kernel to simulate the performance. The product of the simulation is a Gantt chart showing the execution of stars and memory usage over time as well as estimates of certain other system level metrics (weight, size, power, reliability, etc.). This capability has been developed as a front-end architectural trade tool for the Sanders RASSP Program. Contact: Eric Pauer
    A demonstration is available as a patch to Ptolemy0.6.

  • They have developed their own custom code generation domain for FPGAs that uses the DE domain to partition the graph onto multiple FPGA and then automatically insert registers to compensate for pipelining. They apply perl scripts to the resulting ptcl code to generate the FPGA layout in a Xlinx format. Contact: Cory Myers

  • They have integrated Ptolemy into their RASSP Design Environment (RDE), a set of a fifteen CAD tools for hardware/software codesign and thirteen CAD tools for electrical and physical design for embedded signal processors. Reference: Rick Ong, Rob Costantino, and Rodger Philips, "ENvironment and Tools for an Intergrated RDE (ENTIRE)," Proc. ARPA Rapid Prototyping of Application-Specific Signal Processors Conference, pp. 115-121, Arlington, VA, July, 1995. Contact: Rob Costantino

• University of California at Berkeley, Department of Electrical Engineering and Computer Sciences, has used Ptolemy in both research and education.
  • The Infopad Project used Ptolemy as a design, simulation, and test generation tool . The Infopad Project is an indoor wireless communications network consisting of a computer network backbone connecting rooms together, with optical transmitters and hand-held computer-receivers in each room. Simulation of communications protocols between the hand-held terminals and the rest of the Infopad system have been performed in Communicating Processes domain, which is not included in the Ptolemy 0.6 release. Infopad designers also made use of the Synchronous Dataflow (SDF) model of computation to generate standalone demonstrations using the compile-SDF target and the Code Generation in C domain. The primary use of Ptolemy was in designing the hand-held computers, e.g. simulating their behavior, determining bit widths of the internal data paths, and test vector generation. Contact: Sam Sheng

  • Javier Contreras in Prof. Felix Wu's group is using Ptolemy as a framework for studying power systems transmission planning. Using Ptolemy, he is studying tradeoffs in using different optimization techniques provided by library routines and commercial software such as MATLAB. The optimization procedures are chosen graphically using hierarchical block diagrams and interactive Tcl/Tk interfaces. Contact: Javier Contreras

  • Alberto Ferrari in the CAD group is interfacing MatrixX to Ptolemy to leverage their previous implementations of automatic control strategies in MatrixX. Contact: Alberto Ferrari

  • Students in the undergraduate and graduate signal processing courses use Ptolemy as an exploratory laboratory. Contact: Prof. Edward A. Lee

  • Ptolemy was the platform to show real-time signal processing demonstrations in a sophomore class entitled Introduction to Real-Time Digital Systems. Contact: Prof. Edward A. Lee