(MAPLD Conference)
D1: Brad L Hutchings
Brigham Young University, Hewlett Packard Laboratories, Ltd
TBA
Configurable-computing researchers commonly resort to standard ASIC design tools when mapping applications to FPGAs. They have little choice. However, because these tools have been optimized for ASIC design, they often fall short when it comes to achieving high performance designs in FPGAs. When implementing high performance designs with FPGAs, it is usually quite important to control both the structure and flooplan to some degree. Unfortunately, current tools make this more difficult than it needs to be. For example, schematic capture is useful for controlling the low-level structure and placement of designs but is typically useless for creating the parameterized and reusable circuit blocks often found in configurable-computing applications. High-level synthesis tools on the other hand allow for the creation of parameterized designs but often don't provide the designer with sufficient control over the low-level details of a design. Combining these two tools tends to be the only way to achieve reasonable results but the process tends to be time consuming, frustrating, and error prone. In general, we are finding that the shortcomings of these tools hamper designer productivity and ultimately limit the level of performance you can achieve. This talk will discuss the JHDL project, currently underway at Brigham Young University, that seeks to develop a tool suite that overcomes some of the problems found in commercial tools. The tool suite is being developed in parallel with a set of applications and early results are already showing increased productivity. This talk will first introduce the topic by discussing what is unique to configurable-computing applications as opposed to ASIC design and will discuss the various elements of the tool suite, their current status and how they have been tailored specifically for development of configurable-computing applications.Home
Last Revised: July 03, 2002
Digital Engineering Institute
Web Grunt: Richard Katz
