Ardavan Pedram, Andreas Gerstlauer, and Robert Geijn. 2012.
 On the Efficiency of Register File versus Broadcast Interconnect for Collective Communications in Data-Parallel Hardware Accelerators. In <em>Proceedings of the 24th International Symposium on Computer Architecture and High Performance Computing</em>, October 24, 2012, Nova Iorque/EUA, Brasil. SBC, Porto Alegre, Brasil, 19-26.