Getting back to the OP/TS question. We cannot imagine how a single program on a single machine can "optimize" TCP throughput to provide a meaningful improvement. In all probability it seems prudent to consider it just another scam, unless independent evidence of it's efficacy can be provided.