Translating Timing into an Architecture: The Synergy of COTSon and HLS (Domain Expertise—Designing a Computer Architecture via HLS)

Figure 13

Evaluation of the data access latency in the COTSon framework when using the matrix multiplication benchmark by varying cache size, number of nodes of the distributed system, and different Linux distribution (“xenv0” = Ubuntu 16.04, “karmic64” = Ubuntu 9.10, “trusty-axmv3” = Ubuntu 14.04, and “tfxv4” = Ubuntu 10.10). The DSE shows that the data-cache access latency is almost similar in each Linux distribution, but it is lowering when we increase the number of nodes. Thus, multiple-node configuration can be more convenient in the DF-Threads execution model.