MatRox: modular approach for improving data locality in hierarchical (Mat)rix App(Rox)imation

MatRox: modular approach for improving data locality in hierarchical (Mat)rix App(Rox)imation
复制标题

MatRox:用于改进分层 (Mat)rix App(Rox)imation 中数据局部性的模块化方法

DOI:
10.1145/3332466.3374548
复制
发表时间:
2020
期刊:
PPoPP '20: Proceedings of the 25th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming
影响因子:
--
通讯作者:
Dehnavi, Maryam Mehri
Dehnavi, Maryam Mehri
中科院分区:
--
文献类型:
--
作者:
Liu, Bangtian;Cheshmi, Kazem;Soori, Saeed;Strout, Michelle Mills;Dehnavi, Maryam Mehri

文献摘要

参考文献

相似文献

分层矩阵近似在机器学习和科学界获得了巨大的吸引力,因为它们利用了核方法中可用的低阶结构来压缩核矩阵。所得到的压缩矩阵HMatrix用于降低运算的计算复杂度,例如在评估阶段具有可调精度的HMatrix-矩阵乘法。现有的HMatrix求值实现不能保持局部性,经常会导致高同步的不平衡并行执行。此外,如果内核方法或所需精度发生变化,则当前的解决方案需要重新执行压缩阶段。MatRox是一个框架,它使用新颖的结构分析策略和代码专门化和存储格式来改进局部性,并为HMatrix-矩阵乘法创建负载平衡的并行任务。当输入精度和核函数发生变化时,矩阵压缩阶段的模块化使计算得以重复使用。MatRox生成的矩阵-矩阵乘法代码分别比GOFMM、SMASH和STRUMPACK中的库实现快2.98倍、1.60倍和5.98倍。此外,重复使用部分压缩计算来更改精度的能力使MatRox的性能提高了2.64倍,而不是使用GOFMM对精度进行了五次更改。
Hierarchical matrix approximations have gained significant traction in the machine learning and scientific community as they exploit available low-rank structures in kernel methods to compress the kernel matrix. The resulting compressed matrix,HMatrix, is used to reduce the computational complexity of operations such as HMatrix-matrix multiplications with tuneable accuracy in anevaluationphase. Existing implementations of HMatrix evaluations do not preserve locality and often lead to unbalanced parallel execution with high synchronization. Also, current solutions require the compression phase to re-execute if the kernel method or the required accuracy change. MatRox is a framework that uses novel structure analysis strategies with code specialization and a storage format to improve locality and create load-balanced parallel tasks for HMatrix-matrix multiplications. Modularization of the matrix compression phase enables the reuse of computations when there are changes to the input accuracy and the kernel function. The MatRox-generated code for matrix-matrix multiplication is 2.98X, 1.60X, and 5.98X faster than library implementations available in GOFMM, SMASH, and STRUMPACK respectively. Additionally, the ability to reuse portions of the compression computation for changes to the accuracy leads to up to 2.64X improvement with MatRox over five changes to accuracy using GOFMM.
异构系统上的运行时依赖计算和循环执行
DOI: --
发表时间: 2013
期刊: IEEE/ACM International Symposium on Code Generation and Optimization
影响因子: --
作者:
Jayvant Anantpur;R. Govindarajan
通讯作者: R. Govindarajan
用于压缩密集 SPD 矩阵的几何忽略 FMM
DOI: --
发表时间: 2017
期刊: International Conference for High Performance Computing, Networking, Storage and Analysis
影响因子: --
作者:
Chenhan D. Yu;James Levitt;Severin Reiz;G. Biros
通讯作者: G. Biros
几种层次秩结构矩阵算法的稳定性研究
DOI: 10.1137/15m1026195
发表时间: 2016
期刊: SIAM J. Matrix Anal. Appl.
影响因子: --
作者:
Yuanzhe Xi;J. Xia
通讯作者: J. Xia
通用维度中与内核无关的 FMM
DOI: --
发表时间: 2015
期刊: International Conference for High Performance Computing, Networking, Storage and Analysis
影响因子: --
作者:
William B. March;Bo Xiao;Sameer Tharakan;Chenhan D. Yu;G. Biros
通讯作者: G. Biros
DOI: --
发表时间: 2016
影响因子: 3.1
作者:
William B. March;Bo Xiao;Chenhan D. Yu;G. Biros
通讯作者: G. Biros