Title
The design of ultra scalable MPI collective communication on the K computer
Abstract
This paper proposes the design of ultra scalable MPI collective communication for the K computer, which consists of 82,944 computing nodes and is the world's first system over 10 PFLOPS. The nodes are connected by a Tofu interconnect that introduces six dimensional mesh/torus topology. Existing MPI libraries, however, perform poorly on such a direct network system since they assume typical cluster environments. Thus, we design collective algorithms optimized for the K computer.On the design of the algorithms, we place importance on collision-freeness for long messages and low latency for short messages. The long-message algorithms use multiple RDMA network interfaces and consist of neighbor communication in order to gain high bandwidth and avoid message collisions. On the other hand, the short-message algorithms are designed to reduce software overhead, which comes from the number of relaying nodes. The evaluation results on up to 55,296 nodes of the K computer show the new implementation outperforms the existing one for long messages by a factor of 4 to 11 times. It also shows the short-message algorithms complement the long-message ones.
Year
DOI
Venue
2013
10.1007/s00450-012-0211-7
Computer Science - R&D
Keywords
Field
DocType
collective algorithm,long message,ultra scalable,direct network system,long-message algorithm,mpi collective communication,k computer,mpi library,short-message algorithm,multiple rdma network interface,neighbor communication,torus network
Computer science,Grid network,Parallel computing,Collective communication,Software,Remote direct memory access,Latency (engineering),Interconnection,Network interface,Distributed computing,Scalability
Journal
Volume
Issue
ISSN
28
2-3
1865-2042
Citations 
PageRank 
References 
10
0.75
8
Authors
8
Name
Order
Citations
PageRank
Tomoya Adachi1121.84
Naoyuki Shida2334.15
Kenichi Miura3100.75
Shinji Sumimoto451934.67
Atsuya Uno58712.94
Motoyoshi Kurokawa6616.03
Fumiyoshi Shoji7527.36
Mitsuo Yokokawa822751.71