This paper develops and validates an analytical model for evaluating various types of architectural alternatives for shared-memory systems with processors that aggressively exploit instruction-level parallelism. Compared to simulation, the analytical model is many orders of magnitude faster to solve, yielding highly accurate system performance estimates in seconds. The model input parameters characterize the ability of an application to exploit instruction-level parallelism as well as the interaction between the application and the memory system architecture. A trace-driven simulation methodology is developed that allows these parameters to be generated over 100 times faster than with a detailed execution-driven simulator. Finally, this paper shows that the analytical model can be used to gain insights into application performance and to evaluate architectural design trade-offs. 1

TimeBench/SRVN is a prototype toolset for computer-aided design and performance analysis of software, with an emphasis on distributed client-server systems. The performance behaviour of such systems may defy intuition because it involves factors in the software design (such as the partitioning of the functionality and the frequency with which requests will be made to each server) and in the configuration of the distributed system (including replication of services, the distribution of data, and the speed of network access). The novelty of the tool consists in providing support both for developing design specifications and also for performance analysis. The integrated approach avoids the semantic gap between a designer's domain and the performance modeling domain, and assists the designer to explore factors that impact the performance of a design. The performance models are based on the Stochastic Rendezvous Network (SRVN) formalism for client-server systems with synchronous service req...

Early performance estimates for a new software system aid the design process by providing feedback when design decisions can be easily revised. Unfortunately, constructing a performance model of a distributed and concurrent software system can require significant effort. An automated performance model generation technique is described that reduces the model building effort by providing: easy specification of performance experiments, empirical estimates for model parameters, automated model generation, and support for different types of models. A prototype is used to describe a software system, from which causal traces (angio traces) are recorded during execution. These traces are then processed into sequences of resource demands (workthreads), aggregated into system execution descriptions (workthread classes), and combined to generate a performance model. The technique can also be applied at other stages of the development process, including the redesign of existing software. Page ii...

Software Bottlenecks are performance constraints caused by slow execution of a software task. In typical client-server systems a client task must wait in a blocked state for the server task to respond to its requests, so a saturated server will slow down all its clients. A Rendezvous Network generalizes this relationship to multiple layers of servers with send-and-wait interactions (rendezvous), a two-phase model of task behaviour, and to a unified model for hardware and software contention. Software bottlenecks have different symptoms, different behaviour when the system is altered, and a different cure from the conventional bottlenecks seen in queueing network models of computer systems, caused by hardware limits. The differences are due to the "push-back" effect of the rendezvous, which spreads the saturation of a server to its clients. The paper describes software bottlenecks by examples, gives a definition, shows how they can be located and alleviated, and gives a method for estim...

For the development of efficient parallel applications, fast but reliable performance predictions are essential. Many existing modelling formalisms are either not directly suited to model parallel applications, or too expensive. This paper describes several extensions and improvements to a previously introduced methodology, based on an extension of queueing networks. The set of machine model building blocks is extended, a new algorithm for the prediction of multiple-class parallel section completion times is introduced, and it is shown how programs containing conditional statements at the program level and memory hierarchies the machine level are modelled. The concepts introduced in this paper are illustrated by a number of examples throughout the paper, and a case study comparing the predictions to measurements carried out on an actual parallel machine.

. In the development of e#cient parallel applications, reliable performance predictions are essential. However, many performance modelling formalisms, such as queueing networks, are not directly suitable for modelling parallel applications, while for other formalisms the analysis is too expensive. We present a methodology for performance modelling of parallel processing systems #Glamis#, based on extended queueing networks, aiming to overcome these problems. The methodology yields reliable performance predictions for a class of parallel machines and programs at relatively low #polynomial time# analysis cost. Additional reductions of analysis cost are obtained by exploiting inherent replications in parallel systems. 1 Introduction Performance predictions of a parallel program running on a parallel machine can be of great importance in a number of situations. A #rst application is the decision which one from a range of parallel machines is most suitable for a given applicatio...

This report describes the results of preliminary research in the field of probabilistic performance modelling of parallel applications. The work was carried out as part of the ProcMod (Processor Modelling) sub-project of the parTool project. The ProcMod sub-project aims at the development of a performance modelling technique and associated tool support which, based on a generic machine modelling paradigm, predicts parallel application performance at different hierarchical levels. In this way, performance feedback is available at all modelling levels, enabling the use of performance information during all stages of the application development process. The application of this technique is twofold: on the one hand, performance can be optimised by means of feedback to the user (or parallelising compiler) for a given target machine (or machine class); on the other hand, a comparative analysis of target machine performance can be made given a parallel application (or application class). In an earlier report the emphasis was on the description of parallel computer architectures [18]. This report, on the other hand, concentrates on the performance modelling and prediction techniques

In this paper, we develop and apply an analytical model to analyze the performance of the EARTHMANNA multithreaded multiprocessor system. The performance model is based on closed queuing networks. We develop heuristics to account for the realistic subsystem interactions (simultaneous resource possession) and multithreaded workload. Inputs to the analytical model are architectural parameters of the EARTH-MANNA system, and program workload parameters. Predictions of the performance model include processor utilization and network latency. We characterize the variation of these performance measures in terms of program workload and architecture characteristics. We validate the analytical model through runtime measurements from the actual program executions on the EARTH-MANNA system. 1 Introduction Multithreaded architectures have been proposed to tolerate the long latencies for communication and synchronization in parallel program executions [3, 2, 8]. On encountering a long latency opera...

Modelling and analysis of the quantitative performance of business processes plays an important role in different stages of business process re-engineering (BPR). Although simulation is used in most model-based BPRapproaches to obtain performance measures, in many cases analytic performance prediction techniques are preferable. Two aspects which are often distinguished in business process modelling are the (static) organisation structure and the (dynamic) behaviour. Certain hybrid performance analysis methods, originating from the field of parallel computer modelling, combine modelling languages which closely correspond with these two aspects. An additional advantage of hybrid techniques opposed to more traditional performance modelling techniques is that they occupy a favourable position on the trade-off between prediction accuracy and analytical efficiency. We describe such a technique, and show how it can be used for the analysis of business processes. INTRODUCTION Redesign of bus...

As industry moves towards distributed client-server systems, the performance of these systems has gained attention. Distributed systems may consist of a large number of components with functionality distributed throughout the components. Analytical methods, which use queueing theory, are suitable for the performance study of large systems. However, for practical purposes, they may be limited to the size of the system that can be solved. This thesis extends the performance analysis toolset, Stochastic Rendezvous Network (SRVN) models and the Layered Queueing Network Solver (LQNS), to handle large systems with groups of so called replicated components that are homogeneous from a performance modeling point of view. The analytical method described removes the limitation on the size of these large systems that can be analyzed by the toolset. The method includes the use of a simple model definition for large systems and a solver that is quick and memory efficient.