The roadblocks to broad adoption of high level synthesis

High level synthesis provides significant benefit to a few users on a few specific problems, most notably video and audio processing. But after more than a decade of use, many users – as well as critics – state that it is not yet ready for wide-spread adoption. This talk discusses why high level synthesis is limited in its appeal and scope, and what must be done to overcome these limitations. In particular, it describes the problems of source language, verification strategy, and the challenges of describing control logic at a high level. It concludes with some suggestions for bridging the gap between high level synthesis and RTL design, and using this as a strategy for overcoming the major roadblocks to broad adoption of high level synthesis.