As processor development shifts from strict single core frequency scaling to heterogeneous resource scaling two important considerations require evaluation. First, how to design systems with an increasing amount of heterogeneous resources, and second, how to maintain a designer's productivity as the number of possible configurations grows. Therefore, it is necessary to determine what useful information can be gathered from existing designs to help predict or identify a design's potential scalability, as well as, identifying which routine tasks can be automated to improve a designer's productivity. Moreover, once this information is collected, how can this information be conveyed to the designer such that it can be used to increase overall productivity when implementing the design over increasing amounts of resources?

This research looks at various approaches to analyze designs and attempts to distribute an application efficiently across a heterogeneous cluster of computing resources through the use of a Systematic Design Analysis flow and an assortment of productivity tools. These tools provide the designer with projections on the amount of resources needed to scale an existing design to a specified performance, as well as, projecting the performance based on a specified amount of resources. This is accomplished through the combination of static HDL profiling, component synthesis resource utilization, and runtime performance monitoring. For evaluation, four case studies are presented to demonstrate the proposed flow's scalability on a small scale cluster of FPGAs. The results are highly favorable, providing orders of magnitude speedup with minimal intervention from the designer.