Next:
1. Introduction
Up:
thesis
Previous:
List of Figures
 
Contents
List of Tables
2.2.
Breakdown analysis of several lightweight messaging systems with respect to the three-phase data transfer. During the data transfer, we assume that the outgoing message is resided on any virtual address space of the sender process, and is going to be delivered to any virtual address space of the receiver process.
2.3.
Summary of related abstract models which built on the same architectural foundation. With simple message-passing model, the programming interface is based on the send and receive operations, or their variants. While for high-level message-passing model, the programming interface includes simple point-to-point communication as well as complex collective communications.
4.1.
Sample data collected on the 416 platform
4.2.
Comparisons of the congestion behavior observed on the IBM416 and the Cisco2980 switches under the same parameter sets
4.3.
The setting used in emulating the many-to-one flow over a congested uplink port
4.4.
Comparisons of the measured performance on IBM416, Cisco2980 and the IBM uplink port under the same parameter settings.
4.5.
The network configuration used to emulating the multiple one-way data transfer over a congested uplink port
4.6.
The network configuration used to emulating the multiple bi-directional data flow over a congested uplink port
5.1.
Performance characteristics of different Complete Exchange schemes
5.2.
Model parameters for the experiment cluster
6.1.
The
parameter of different switches in our experimental setup
A.1.
parameters of different Ethernet switches