Optimised gather collectives on QsNetII

Authors:
Duncan Roweth;David Addison
Affiliations:
Quadrics Ltd, Bristol, United Kingdom;Quadrics Ltd, Bristol, United Kingdom
Venue:
PVM/MPI'05 Proceedings of the 12th European PVM/MPI users' group conference on Recent Advances in Parallel Virtual Machine and Message Passing Interface
Year:
2005

Citing 3
Cited 1

MPI-The Complete Reference, Volume 1: The MPI Core

MPI-The Complete Reference, Volume 1: The MPI Core
Hardware- and Software-Based Collective Communication on the Quadrics Network

NCA '01 Proceedings of the IEEE International Symposium on Network Computing and Applications (NCA'01)
The Case of the Missing Supercomputer Performance: Achieving Optimal Performance on the 8,192 Processors of ASCI Q

Proceedings of the 2003 ACM/IEEE conference on Supercomputing

Efficient shared memory and RDMA based collectives on multi-rail QsNetII SMP clusters

Cluster Computing

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we describe the implementation of the gather and allgather collectives on QsNetII. Results from a cluster of 980 4-CPU nodes show good latencies, bandwidths and scaling, with a 3920 process, 8-byte, gather completing in 88 microsecs.