Communication-Avoiding Cholesky-QR2 for Rectangular Matrices

Communication-Avoiding Cholesky-QR2 for Rectangular Matrices
复制标题

矩形矩阵的通信避免 Cholesky-QR2

DOI:
10.1109/ipdps.2019.00020
复制
发表时间:
2017
期刊:
2019 IEEE International Parallel and Distributed Processing Symposium (IPDPS)
影响因子:
--
通讯作者:
Edgar Solomonik
Edgar Solomonik
中科院分区:
--
文献类型:
--
作者:
Edward Hutter;Edgar Solomonik

文献摘要

参考文献

被引文献

相似文献

可扩展的QR分解算法求解最小二乘和特征值问题是至关重要的,因为现代机器中的并行性越来越高。我们介绍了一个更一般的并行化的CholeskyQR 2算法,并显示其有效性的矩阵大小的范围很广。我们的算法在3D处理器网格上执行,其尺寸可以调整到权衡同步,处理器间通信,计算工作和内存占用的成本。我们实现了这个算法,产生一个代码,可以实现一个因子的Θ(P^1/6)少P处理器上的处理器间通信比任何以前的并行QR实现。我们在英特尔Knights-Landing和Cray XE超级计算机上的性能研究证明了这种CholeskyQR 2并行化在大量节点上的有效性。具体来说,相对于ScaLAPACK的QR,在Stampede 2的1024个节点上,我们的CholeskyQR 2实现在强缩放测试中快了2.6 - 3.3倍,在弱缩放测试中快了1.1 - 1.9倍。
Scalable QR factorization algorithms for solving least squares and eigenvalue problems are critical given the increasing parallelism within modern machines. We introduce a more general parallelization of the CholeskyQR2 algorithm and show its effectiveness for a wide range of matrix sizes. Our algorithm executes over a 3D processor grid, the dimensions of which can be tuned to trade-off costs in synchronization, interprocessor communication, computational work, and memory footprint. We implement this algorithm, yielding a code that can achieve a factor of Θ(P^1/6) less interprocessor communication on P processors than any previous parallel QR implementation. Our performance study on Intel Knights-Landing and Cray XE supercomputers demonstrates the effectiveness of this CholeskyQR2 parallelization on a large number of nodes. Specifically, relative to ScaLAPACK's QR, on 1024 nodes of Stampede2, our CholeskyQR2 implementation is faster by 2.6x-3.3x in strong scaling tests and by 1.1x-1.9x in weak scaling tests.
Stampede 2:XSEDE 超级计算机的演变
DOI: 10.1145/3093338.3093385
发表时间: 2017
期刊: Success and Impact (PEARC17
影响因子: --
作者:
Stanzione, Dan;Barth, Bill;Gaffney, Niall;Gaither, Kelly;Hempel, Chris;Minyard, Tommy;Mehringer, S.;Wernert, Eric;Tufo, H.;Panda, D.
通讯作者: Panda, D.