Scalability of Hybrid Sparse Matrix Dense Vector (SpMV) Multiplication
Scalability of Hybrid Sparse Matrix Dense Vector (SpMV) Multiplication
复制标题
DOI：
10.1109/hpcs.2018.00072
复制
发表时间：
2018-07
期刊：
影响因子：
--
通讯作者：
Brian A. Page;P. Kogge
中科院分区：
文献类型：
--
作者：
Brian A. Page;P. Kogge
SPMV是稀疏矩阵和密集向量的乘积，它象征着新的类别的应用程序带宽和通信的应用程序，而不是flop，驱动的。此类计算中的稀疏性和随机性会破坏常规的实现，尤其是在强大而不是弱的情况下尝试缩放。本文研究改进了具有更好性能的混合SPMV代码，尤其是对于此类问题最稀少的。数据放置和远程减少的问题都是在一系列矩阵特征上建模的。那些限制强可伸缩性的因素被量化。
SpMV, the product of a sparse matrix and a dense vector, is emblematic of a new class of applications that are memory bandwidth and communication, not flop, driven. Sparsity and randomness in such computations play havoc with conventional implementations, especially when strong, instead of weak, scaling is attempted. This paper studies improved hybrid SpMV codes that have better performance, especially for the sparsest of such problems. Issues with both data placement and remote reductions are modeled over a range of matrix characteristics. Those factors that limit strong scalability are quantified.