Multicore-aware parallel temporal blocking of stencil codes for shared and distributed memory
Multicore-aware parallel temporal blocking of stencil codes for shared and distributed memory
复制标题
用于共享和分布式内存的多核感知并行模板代码时间块
DOI：
--
复制
发表时间：
2009
期刊：
影响因子：
--
通讯作者：
G. Wellein
中科院分区：
文献类型：
--
作者：
M. Wittmann;G. Hager;G. Wellein
需要新的算法和优化技术来平衡带宽占主导地位的多功能芯片的加速趋势。众所周知，可以通过时间阻挡来改善模板代码的性能，从而减少内存界面的压力。我们介绍了一种新的管道方法，该方法可以在多核心环境中明确使用共享的缓存，并最大程度地减少同步和边界开销。对于共享内存节点的簇，我们证明了如何成功地在混合共享/分布式内存环境中成功使用时间阻滞。
New algorithms and optimization techniques are needed to balance the accelerating trend towards bandwidth-starved multicore chips. It is well known that the performance of stencil codes can be improved by temporal blocking, lessening the pressure on the memory interface. We introduce a new pipelined approach that makes explicit use of shared caches in multicore environments and minimizes synchronization and boundary overhead. For clusters of shared-memory nodes we demonstrate how temporal blocking can be employed successfully in a hybrid shared/distributed-memory environment.