Data Sharing Mechanisms for Parallel Graph Algorithms on the Intel SCC

Abstract—On many-core processors that do not provide hard-ware cache coherence, using shared memory in parallel com-putations is challenging. Reverting to pure message passing would avoid consistency issues, but replicating large shared datasets by messages is less efficient than accessing them di-rectly through shared memory. The TACO-MESH framework provides lightweight remote method calls and shared objects with software-managed consistency. This paper presents experience from porting a graph partitioning algorithm to the framework. A performance evaluation on the experimental Intel SCC pro-cessor, which has no hardware cache coherence, shows that parallelization can be efficient despite the overhead of software-level consistency management. Index Terms—many-core, shared memory, cache coherence, graph partitioning I