Design and Implementation of Multiple Fault-Tolerant MPI over Myrinet Session: MPI and Network Transport Event Type: Paper Time: 11:30am - 12:00pm Session Chair: Scott Pakin Speaker(s): Hyungsoo Jung, Dongin Shin, Hyuck Han, Jai W. Kim, Heon Y. Yeom, Jongsuk Lee Location: 608-609 Abstract: Advances in network technology and computing power have inspired the emergence of high-performance cluster computing systems. While cluster management and harware high-availability tools are readily available, practical and easily deployable fault-tolerant systems have not been successfully adopted commercially. We present a fault-tolerant system, Multiple fault-tolerant MPI over Myrinet (M-cube), that differs in notable respects from other proposed fault-tolerant systems in the literature. M-cube is built on top of Myrinet since it is regarded as one of the best solutions for high-performance networks and is widely used in cluster computing systems. The features of M-cube are that it requires no modifications of application code, and it preserves much of the high performance characteristics of Myrinet. Experimental results substantiate our assertion that M-cube can be a good candidate for practically deployable fault-tolerant systems in a very large and high performance Myrinet cluster, and its protocol can be applied to a wide variety of parallel communication libraries without difficulty. This paper can be found in the ACM and IEEE Digital Libaries Click here for ACM Click here for IEEE Chair/Speaker Details: Scott Pakin (Chair) Los Alamos National Laboratory Hyungsoo Jung Seoul National University Dongin Shin Seoul National University Hyuck Han Seoul National University Jai W. Kim Seoul National University Heon Y. Yeom Seoul National University Jongsuk Lee Korea Institute of Science and Technology Information