The impact of MPI queue usage on message latency
- 1 January 2004
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 8 (01903918) , 152-160 vol.1
- https://doi.org/10.1109/icpp.2004.1327915
Abstract
It is well known that traditional microbenchmarks do not fully capture the salient architectural features that impact application performance. Even worse, microbenchmarks that target MPI and the communications subsystem do not accurately represent the way that applications use MPI. For example, traditional MPI latency benchmarks time a ping-pong communication with one send and one receive on each of two nodes. The time to post the receive is never counted as part of the latency. This scenario is not even marginally representative of most applications. Two new microbenchmarks are presented here that analyze network latency in a way that more realistically represents the way that MPI is typically used. These benchmarks are used to evaluate modern high-performance networks, including Quadrics, InfiniBand, and Myrinet.Keywords
This publication has 11 references indexed in Scilit:
- An analysis of NIC resource usage for offloading MPIPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2004
- An evaluation of current high-performance networksPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2004
- A New MPI Implementation for Cray SHMEMPublished by Springer Nature ,2004
- Performance Comparison of MPI Implementations over InfiniBand, Myrinet and QuadricsPublished by Association for Computing Machinery (ACM) ,2003
- High performance RDMA-based MPI implementation over InfiniBandPublished by Association for Computing Machinery (ACM) ,2003
- Measuring MPI Latency VariancePublished by Springer Nature ,2003
- The Quadrics network: high-performance clustering technologyIEEE Micro, 2002
- LogGP: Incorporating Long Messages into the LogP Model for Parallel ComputationJournal of Parallel and Distributed Computing, 1997
- Effects of communication latency, overhead, and bandwidth in a cluster architectureACM SIGARCH Computer Architecture News, 1997
- Myrinet: a gigabit-per-second local area networkIEEE Micro, 1995