Today ISC 2016 announced that a research paper in the area of Message Passing Interface (MPI) performance, has been selected to receive the 2016 Hans Meuer Award. The awarding will take place at the ISC High Performance conference on Monday, June 20.
Titled “Mitigating MPI Message Matching Misery,” the paper was submitted by Mario Flajslik, James Dinan, and Keith Underwood of Intel Corporation. In it, they propose a series of structures and algorithms to address the performance challenges associated with MPI message ordering in the presence of wildcards. Performance problems arise when applications scale up, since the traditional approach of using linked lists requires a lot of overhead. Their solution is to replace this with a hashing approach, along with using additional metadata to greatly speed up MPI message matching. Using a number of different benchmarks, they were able to demonstrate significant performance increases, while maintaining MPI ordering semantics.
James Dinan will present the winning paper on his team’s behalf during the research paper session on Monday afternoon. This paper was selected out of a total of 25 accepted submissions by the research paper committee, which is led by Prof. Dr. Jack Dongarra this year.
The winners will receive a cash prize of 3,000 euros, an award certificate, and a free conference pass for ISC 2017. The Hans Meuer Award has been introduced in memory of the late Dr. Hans Meuer, general chair of the ISC conference from 1986 through 2014, and co-founder of the TOP500 project.
Reacting to his team’s success, Dinan said, “My colleagues and I are honored that our submission was selected over 24 other papers for the Hans Meuer Award.” On their research findings, he said, “reducing communication latency is extremely difficult, and work like this that revisits basic assumptions is crucial to making headway on this important challenge.”
ISC has posted a complete list of papers that will be presented at the 2016 Research Paper session.
Register now for ISC 2016 to enjoy Early Bird rates.
