The Community for Technology Leaders
Proceedings of the 22nd International Conference on Parallel Architectures and Compilation Techniques (2007)
Brasov, Romania
Sept. 15, 2007 to Sept. 19, 2007
ISSN: 1089-795X
ISBN: 0-7695-2944-5
pp: 270-279
Xiaofeng Guo , Google Inc
Jinquan Dai , Intel China Software Center, China
Long Li , Intel China Software Center, China
Zhiyuan Lv , Intel China Software Center, China
Prashant R. Chandra , Intel Corporation
ABSTRACT
Network processors employ a multithreaded, chip-multiprocessing architecture to effectively hide memory latency and deliver high performance for packet processing applications. In such a parallel paradigm, when multiple threads modify a shared variable in the external memory, the threads should be properly synchronized such that the accesses to the shared variable are protected by critical sections. Therefore, in order to efficiently harness the performance potential of network processors, it is critical to hide the memory latency and synchronization latency in multi-threading and multiprocessing. In this paper, we present a novel program transformation used in the Intel? Auto-partitioning C Compiler for IXP, which perform optimal placement of memory access instructions and synchronization instructions for effective latency hiding. Experimental results show that the transformation provides impressive speedup (up-to to 8.5x) and scalability (upto 72 threads) of the performance for the real-world network application (a 10Gbps Ethernet Core/Metro Router).
INDEX TERMS
null
CITATION
Xiaofeng Guo, Jinquan Dai, Long Li, Zhiyuan Lv, Prashant R. Chandra, "Latency Hiding in Multi-Threading and Multi-Processing of Network Applications", Proceedings of the 22nd International Conference on Parallel Architectures and Compilation Techniques, vol. 00, no. , pp. 270-279, 2007, doi:10.1109/PACT.2007.46
92 ms
(Ver 3.3 (11022016))