DiMP: Architectural Support for Direct Message Passing on Shared Memory Multi-cores

Thanks to programming approaches like actor-based models, message passing is regaining popularity outside large-scale scientific computing for building scalable distributed applications in many-core processors. Unfortunately, the mismatch between message passing models and today´s shared-memory hardware provided by commercial vendors results in suboptimal performance and loss of efficiency. This paper presents a set of architectural extensions to reduce the overheads incurred by message passing workloads running on shared memory multi-core architectures. It describes the instruction set extensions and the hardware implementation. In order to facilitate programmability, the proposed extensions are used by a message passing library, allowing programs to take advantage of them transparently. As a proof-of-concept, we use a modified MPICH library and MPI programs to evaluate the proposal. Experimental results show that, on average, our proposal spends 60% less cycles performing data transfers in MPI functions, and reduces the L1 data cache misses in said functions to a fourth.