Algorithm Analysis and Mapping Environment for Adaptive

We are developing an integrated algorithm analysis and mapping environment particularly tailored for signal processing applications on Adaptive Computing Systems (ACS). Our environment allows a designer to map signal processing algorithms to an ACS faster, by an order of magnitude, than is currently possible. Our approach has been to focus on three areas of capability critical to the success of adaptive computing and to integrate these capabilities into an open, extensible software framework [1]. The development of the three areas, algorithm analysis, algorithm mapping, and smart generators, are taking advantage of the special characteristics of signal processing algorithms to reduce the time to eld the ACS. Figure 1 shows a conceptual view of our environment. These capabilities are being implemented as extensions to the Ptolemy design environment developed at the University of California, Berkeley [2]. Algorithm implementation for ACS requires careful consideration of the appropriate signal representation and the costs of operations. The algorithm analysis capabilities being developed on this program will reduce the e ort required to nd good ACS implementation choices for a signal processing algorithm. The environment will provide algorithm designers with information about operation counts, including adds, multiplies, and memory accesses, and with analyses of quantization e ects related to ACS implementations. For many DSP problems, reduced precision arithmetic will maintain acceptable system performance. A mapping of an algorithm to an FPGA architecture will be successful if the designer can limit wordlength growth without sacri cing algorithm performance. Wordlength reduction introduces noise into the data stream, so the designer must balance the need for an e cient implementation with output quality. Our environment supports both analytical and simulationbased wordlength optimization. With these capabiliFloating Point Simulation