Benchmarking

It would be nice, and a natural part of developing various decoding algorithms, to provide speed and performance functions, including plotting.

Computer Science people care only about whether a (hard-decision) decoding was completely succesful or not. In this case, benchmarking is easy: performance is simply a percentage of success. The code parameters are usually supplied by the application and cannot be chosen.

For half-minimum-distance decoders or list decoders, success should be 100% when not too many errors occur.

Engineers can choose between any existing codes, and to compare multiple codes, one plots SNB-vs-Failure Probability (waterfall plots). Probability of failure can then be measured as % of correct blocks, as above, or as % of correct bits, depending on whatever is suitable to the application (or puts the proposed code in the best light).