In message passing environments, the message send time is dominated by overheads that are relatively independent of the message size. Therefore, fine-grained applications (such as Time-Warp simulators) suffer high overheads because of frequent communication. In this paper, we investigate the optimization of the communication subsystem of Time-Warp simulators using dynamic message aggregation. Under this scheme, Time-Warp messages with the same destination LP, occuring in close temporal proximity are dynamically aggregated and sent as a single physical message. Several aggregation strategies that attempt to minimize the communication overhead without harming the progress of the simulation (because of messages being delayed) are developed. The performance of the strategies is evaluated for a network of workstations, and an SMP, using a number of applications that have different communication behavior. 1 Introduction In distributed environments the performance of the communication subsy...

Message Passing Interface (MPI) is an attempt to standardize the communication library for distributed-memory computing systems. Since the release of the recent MPI specification, several MPI implementations have been made publicly available. Different implementations employ different approaches, and thus, the performance of each implementation may vary. Since the performance of communication is extremely crucial to message-passing based applications, selecting an appropriate MPI implementation becomes critical. Our study is intended to provide a guideline on how to perform such a task on workstation clusters which are known to be an economical and effective platform in high performance computing. We investigate several MPI aspects including its functionalities and performance. Our results also point out the strength and weakness of each implementation on our experimental system. 1 Introduction Message Passing Interface (MPI) has become an emerging standard for implementing message-b...

Users of parallel machines need to have a good grasp for how different communication patterns and styles affect the performance of message-passing applications. LogGP is a simple performance model that reflects the most important parameters required to estimate the communication performance of parallel computers. The message passing interface (MPI) standard provides new opportunities for developing high performance parallel and distributed applications. In this paper, we use LogGP as a conceptual framework for evaluating the performance of MPI communications on three platforms: Cray-Research T3D, Convex Exemplar 1600SP, and a network of workstations (NOW). Our objective is to identify a performance model suitable for MPI performance characterization and to compare the performance of MPI communications on several platforms.

Communication is expensive and inevitable in distributed environments. In distributed applications, the speedup achieved by exploiting spatial parallelism and concurrency is dependent on several factors such as communication overhead, message latency, and the number of messages communicated. There have been different approaches to reduce the aforementioned factors and hence the cost associated with communication operations. This thesis explores runtime mechanisms that reduces the communication overhead of distributed applications. It studies the aggregation of messages i.e., collecting application messages in close temporal and spatial proximity to reduce the overall overhead involved in sending and receiving messages. The aggregation mechanism called Dynamic Message Aggregation (DyMA), performs runtime decisions and has been proposed for the asynchronous message passing applications. Three aggregation strategies have been proposed and studied. Two strategies, namely Fixed Aggregation Window (FAW) and Simple Adaptive Aggregation Window (SAAW), are windowing schemes that aggregate messages for a specified time window and sends them as a single physical message. The third strategy namely Fixed Message Count (FMC) aggregate messages for a specified number of message count and sends them as a single physical message. As a case study, the aggregation strategies have been designed and studied for the Time Warp simulation paradigm. DyMA strategies reduce the overall communication overhead and is minimally intrusive with the progress of the simulation. Message aggregation strategies incorporated at the communication layer of asynchronous distributed applications can be employed with other application optimizations or hardware specific communication optimizations. To Amma, Appa...