Montly R&D Status Report Title: "SYSTEM-LEVEL DESIGN METHODOLOGY FOR EMBEDDED SIGNAL PROCESSORS" Contract Number: F33615-93-C-1317 Principal Investigator: Edward A. Lee Organization: University of California at Berkeley 1. Tasks Performed We have ramped up immediately to the staffing level that we proposed. In professional staff, Christopher Hylands has joined the group as a system administrator. Christopher was previously system administrator for the microfabrication lab here at Berkeley. Brian Evans has joined the group as a postdoctoral researcher. Brian just completed his Ph.D. under the direction of Prof. J. McClellan at Georgia Institute of Technology. Brian's background and interests match most aspects of the proposed work. Alan Kamas continues with the group, but is now spending 60% of his effort on this project. Alan is in charge of software coordination, documentation, releases, and support. Mary Stewart continues with the group as grant administrator, but now spending 60% of her effort on this project. Because of careful planning for the initiation of this project, our first month on the project has already produced some interesting results, detailed below. 2. Significant Accomplishments In regard to (3.6) of the SOW, "Interactive, direct-manipulation graphical user interfaces", we have connected a Tcl kernel and its associated toolkit Tk to the Ptolemy kernel. Work is under way to define a set of Tcl procedures for interacting with the kernel. A prototype of an interactive run control panel for simulations has been constructed. Early versions of other graphical interfaces have also been constructed. In regard to (3.7), "Heterogeneous Scheduling", we have used the Tcl interface to build an early prototype of a graphical partitioning tool that a designer can use to direct subsystem designs towards hardware synthesis, software synthesis, or simulation. We have also developed a proposal for combining our previously developed loop scheduling (used for synthesizing compact software implementing multirate DSP systems) with parallel scheduling. In regard to (3.8) of the SOW, "Formal methods", we have developed a mathematical test for synchronous dataflow graphs that determines whether a finite blocking factor can achieve maximal parallelism in a blocked schedule. We are not sure whether this result is novel (a detailed literature search is under way), but we are sure it is useful; we expect to incorporate it into parallel schedulers in order to enhance the parallelism in algorithms. 3. Problems Encountered Our plan to Hire Joe Buck fell through because Joe was given a very attractive offer by Synopsys. However, the net effect of this may ultimately be extremely positive. In hiring Joe, Synopsys has shown considerable interest in the commercialization possibilities of Ptolemy. We expect, therefore, that Joe will continue contributing to our overall objectives, possibly with considerable additional investment from Synopsys. Moreover, because he did not accept our offer, we were able to make an offer to Brian Evans. Brian will bring a fresh perspective and a great deal of experience to the project. 4. Schedule Reconciliation We are on schedule. 5. Next Period Activities The library of Tcl procedures for interacting with the Ptolemy kernel will continue to be developed. The library of Tk widgets replacing the old MIT Athena widgets in the Ptolemy Graphical Interface will also continue to be developed. The proposed parallel loop scheduler will be implemented over the next few months. The blocking factor result mentioned above will be documented, and we will investigate its implications for our existing parallel schedulers. In connection with (3.9), "Software Distribution", planning will begin for our next software release (to be numbered 0.5). 6. Budget Summary Our expenditures may be 15 days behind what we planned due to lags in transitioning new personnel to this project. Details will be provided by the University accounting office. 7. Conferences and Trips The P.I. attended part of an coordination meeting by Martin Marietta, one of the RASSP prime contractors, on October 12. The purpose for this visit was to explain the philosophy of hierarchical heterogeneity as implemented in the Ptolemy kernel. The P.I. also made a presentation at the Infopad Meeting on October 14, and the group conducted software demonstrations on the 15th. This meeting was an industrial participation meeting in connection with another ARPA project at Berkeley, directed by Bob Parker at ARPA. This project uses Ptolemy extensively, providing valuable feedback and application development. The kickoff meeting has been scheduled for October 20 and 21 in Washington DC, slightly beyond the 30 days stipulated by the contract. 8. Other Comments We welcome feedback on the content and format of this report, as we would like these reports as useful as possible.