Showing papers by "Gregory D. Peterson published in 1994"

Beyond execution time: expanding the use of performance models

Abstract: Improved performance is a major motivation for using parallel computation. However, performance models are frequently used only to predict an algorithm's execution time, not to accurately evaluate how the choices of architecture, operating system, interprocessor communication protocol, and programming language also dramatically affect parallel performance. We have developed an analytic model for synchronous iterative algorithms running on distributed-memory MIMD machines, and refined it for disrete-event simulation. The model describes the execution time of a single run in terms of application parameters such as the number of iterations and the required computation in each, and architectural parameters such as the number of processors, processor speed, and communication time. Our experience has shown us that an analytic model can not only accurately predict an algorithm's performance but can also match the algorithm to an appropriate architecture, identify ways to improve the algorithm's performance, quantify the performance effects of algorithmic or architectural changes, and provide a better understanding of how the algorithm works. >

Parallel application performance on shared, heterogeneous workstations

Abstract: The utilization of networked, shared, heterogeneous workstations as an inexpensive parallel computational platform is an appealing idea. However, most performance models for parallel computation are oriented towards the use of tightly-coupled, dedicated, homogeneous processors. We develop and validate an analytic performance modeling methodology for synchronous iterative algorithms executing on networked workstations. The model includes the effects of application load, background load, and processor heterogeneity. We use two applications, nonlinear optimization and discrete-event simulations to validate the model with homogeneous and heterogeneous workstation networks. While generating synthetic background load, we perform validation experiments under various loading conditions. The performance modeling methodology proves to be quite accurate in characterizing the effects of application and background load imbalance. The performance modeling methodology enables us to find the optimal set of processors to use for a synchronous iterative application running on a network, which is important given the emergence of systems that service batches of applications on networked resources. Various policies for the use of the workstations are then considered and the optimal (or near-optimal) scheduling found. Policy choices we investigate include determining the computational costs for individual workstations and relating the priority of background load usage versus the parallel application. We also consider basing costs on time, loading conditions, or other factors. We discuss related problems, such as load balancing, for improving the efficiency with which we utilize networked resources. The accurate performance modelling methodology provides significant help in addressing these and similar issues.

Sharing networked workstations: a performance model

Abstract: The utilization of networked, shared, heterogeneous workstations as an inexpensive parallel computational platform is an appealing idea. However, most performance models for parallel computation are oriented towards the use of tightly-coupled, dedicated, homogeneous processors. We develop and validate an analytic performance model for synchronous iterative algorithms executing on networked workstations. The model includes the effects of application load, background load, and processor heterogeneity. The model is validated using a pair of applications: a nonlinear optimization code and discrete-event simulation. >