In this paper we describe our experience in pursuing two implementation strategies for the BBS communication benchmarks on the Parsys SN1000 SuperNode architecture. We detail the benchmarks themselves and explain the way in which they measure various aspects of communication performance. We give an account of the effort involved in porting the benchmarks to a model of parallelism noticeably different from that in which they were conceived. We present the results of the two strategies, derive some performance factors and discuss the issues and trade offs that arise in the comparison of the two strategies. Additionally, we describe how conducting the benchmarks lead to improvements to the SN1000 operating software through the evolution of a portable design of parallel application loader.