TY - GEN
T1 - Data locality aware strategy for two-phase collective I/O
AU - Filgueira, Rosa
AU - Singh, David E.
AU - Pichel, Juan C.
AU - Isaila, Florin
AU - Carretero, Jesús
N1 - Funding Information:
especially Shiwen Liu, Xiaowen Liu, and Mengjie Du, for their help in the literature survey and drawing; Yanfeng Shi for editing; and Guoxian Gao for the visualization of some computational results. We would also like to thank Prof. Ying Hu of East China University of Science Technology for illuminative discussions on the theoretical background of the EMMS model. We are grateful to the financial support from the National Natural Science Foundation of China (grants 21225628, 21206167, 91434113, 91434201, 21106155, and 2110645), the CAS (grants XDA07080000 and QYZDB-SSW-SYS029 and CAS Interdisciplinary Innovation Team), and the National Program on Key Basic Research Project (grant 2015CB251402).
Funding Information:
The authors would like to thank all members of the EMMS group for their contributions to the studies and developments reviewed in this article and for their collaboration and support in its preparation, especially Shiwen Liu, Xiaowen Liu, and Mengjie Du, for their help in the literature survey and drawing; Yanfeng Shi for editing; and Guoxian Gao for the visualization of some computational results. We would also like to thank Prof. Ying Hu of East China University of Science Technology for illuminative discussions on the theoretical background of the EMMS model. We are grateful to the financial support from the National Natural Science Foundation of China (grants 21225628, 21206167, 91434113, 91434201, 21106155, and 2110645), the CAS (grants XDA07080000 and QYZDBSSW-SYS029 and CAS Interdisciplinary Innovation Team), and the National Program on Key Basic Research Project (grant 2015CB251402).
PY - 2008
Y1 - 2008
N2 - This paper presents Locality-Aware Two-Phase (LATP) I/O, an optimization of the Two-Phase collective I/O technique from ROMIO, the most popular MPI-IO implementation. In order to increase the locality of the file accesses, LATP employs the Linear Assignment Problem (LAP) for finding an optimal distribution of data to processes, an aspect that is not considered in the original technique. This assignment is based on the local data that each process stores and has as main purpose the reduction of the number of communication involved in the I/O collective operation and, therefore, the improvement of the global execution time. Compared with Two-Phase I/O, LATP I/O obtains important improvements in most of the considered scenarios.
AB - This paper presents Locality-Aware Two-Phase (LATP) I/O, an optimization of the Two-Phase collective I/O technique from ROMIO, the most popular MPI-IO implementation. In order to increase the locality of the file accesses, LATP employs the Linear Assignment Problem (LAP) for finding an optimal distribution of data to processes, an aspect that is not considered in the original technique. This assignment is based on the local data that each process stores and has as main purpose the reduction of the number of communication involved in the I/O collective operation and, therefore, the improvement of the global execution time. Compared with Two-Phase I/O, LATP I/O obtains important improvements in most of the considered scenarios.
UR - http://www.scopus.com/inward/record.url?scp=58349099835&partnerID=8YFLogxK
U2 - 10.1007/978-3-540-92859-1_14
DO - 10.1007/978-3-540-92859-1_14
M3 - Conference contribution
AN - SCOPUS:58349099835
SN - 3540928588
SN - 9783540928584
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 137
EP - 149
BT - High Performance Computing for Computational Science - VECPAR 2008 - 8th International Conference, Revised Selected Papers
T2 - 8th International Conference on High Performance Computing for Computational Science, VECPAR 2008
Y2 - 24 June 2008 through 27 June 2008
ER -