Applicability

In general, the presented model can be applied to any kind of instruction processing, where a single instruction is composed of microinstructions. Obviously, the model doesn't produce good results, if there is a strict order of those microinstructions, since in this case no parallel execution of microinstructions can occur.

Overlapping execution of instructions comes automatically with this model. Thus, it can best be applied if dependencies between consecutive instructions are minimized.

The great advantage of this model is that the execution of an instruction is not dependent on the exact timing of FUs. Thus, FUs can be replaced at runtime with other versions of different characterizations. The same holds for the communication structure, which can be adapted to the requirements of the running applications. Thus, this model allows us to optimize global goals like performance or energy consumption.

As previously mentioned, intermediate virtual assembly languages like Java bytecode or the .NET code seem to be good candidates for instruction sets. The range of FU implementations and communication structures is especially wide, if the instruction set has a very high abstraction level and basic operations are sufficiently complex.

Finally, the data driven approach makes it possible to easily integrate new FUs and create new instructions to use these FUs.