As “open source” C/C++ algorithms become an increasingly popular alternative to royalty-based code in embedded processing applications, they bring new technical challenges. Foremost among these is how to optimize the acquired code to work well on the chosen processor. This issue is paramount because a compiler written for a given processor family will exploit that processor’s strengths at the possible expense of inefficiencies in other areas. Performance can be degraded when the same algorithm is run directly out-of-the-box on a different platform. This article will explore the porting of such open-source algorithms to Analog Devices Blackfin processors, outlining in the process a “plan of attack” leading to code optimization.

Reproduced with the permission of Analog Devices, Inc.