Memory-mapping support for reducer hyperobjects

hyperobjects (reducers) provide a linguistic abstraction for dynamic multithreading that allows different branches of a parallel program to maintain coordinated local views of the same nonlocal variable. In this paper, we investigate how thread-local memory mapping (TLMM) can be used to improve the performance of reducers. Existing concurrency platforms that support reducer hyperobjects, such as Intel Cilk Plus and Cilk++, take a hypermap approach in which a hash table is used to map reducer objects to their local views. The overhead of the hash table is costly — roughly 12× overhead compared to a normal L1-cache memory access on an AMD Opteron 8354. We replaced the Intel Cilk Plus runtime system with our own Cilk-M runtime system which uses TLMM to implement a reducer mechanism that supports a reducer lookup using only two memory accesses and a predictable branch, which is roughly a 3× overhead compared to an ordinary L1-cache memory access. An empirical evaluation shows that the...

• Type: Journal
• Year: 2012
• Where: SPAA
• Authors: I.-Ting Angelina Lee, Aamir Shafi, Charles E. Leiserson
• Keywords: AMD Opteron, Distributed And Parallel Computing, Dynamic Multithreading, Memory Accesses, SPAA 2012