D. Ferrari. Computer Systems Performance Evaluation. Prentice Hall, 1978.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... response to changes in workload [13] Lucas conveys a similar message encouraging system designers to regularly run benchmarks and build in as much instrumentation as possible [11] While many systems of the 70 s heeded these words of wisdom and built in significant performance evaluation tools [8], today s systems show a surprising dearth of native mea surement tools. The 1980 s produced a noticeable absence of wellinstrumented systems. Today s common research platform, UNIX, initially had very little in the way of perfor mance measurement tools, but now has a standard set of utilities ....

Ferrari, D., Computer Systems Performance Evaluation, Prentice Hall, Englewood Cliffs NJ, 1978, 44- 64.

....the evaluation and modelling of many relevant performance and reliability indices of probabilistic systems, where one commonly refers to the long run average time required by the system to perform given tasks. By long run average it is therefore meant an average over an in nite period of time [10, 16]. On the other hand, the average behaviour of a probabilistic program can also be captured by considering the average among all its possible results from a given initial store. We will show that this correspond to the expected value of a suitably de ned random variable expressing the particular ....

D. Ferrari. Computer Systems Performance Evaluation. Prentice-Hall, 1978.

....the global age. This metric is computed by taking the average of the differences between the version of the node performing the read and the versions of its neighbours, parent and children. In the experiments, the means of the desired measurements are obtained by using the methods of batch means [Fer78] Num batch batches have been done for each run. Each run is left to run until at least batch length jobs are generated at each node in the network. To eliminate the warmup effect, after generating warmup per jobs at every node, all the statistics are reset. The results reported here are within ....

D. Ferrari. Computer systems Performance Evaluation. Prentice Hall, 1978.

....to 100 CPU utilization. The virtual memory sys 57 tem had between 30 to 50 megabytes free at all times. There were no pageouts, process swaps, or page scans. The file system was finely tuned, with load being well bal anced between the three disks. We obtained trace data by placing cheekpoints [3] at the entry and exit points of evew system call that might potentially cause any file system operation, and within all major routines which affect file system related data structures. Each trace event was timestamped by a hardware clock providing a resolution of 2 milliseconds. This allowed us ....

Ferrari, D., "Computer Systems Performance Evaluation.", Prentice Hall, 1981.

.... management (a) Performance Analysis (b) Performance Management Figure1: From Performance Analysis to Performance Management The traditional approachtoevaluate the performance of computer systems and networks is an off line performance analysis as depicted in Figure 1 (a) see for example [10, 20, 7]) Starting with a characterisation of the system under study and a characterisation of the load, a performance model is built and performance results are obtained by applying performance evaluation techniques (including analytical, numerical, or simulaton techniques) Alternatively, performance ....

D. Ferrari. Computer Systems Performance Evaluation. Prentice-Hall, Englewood Cliffs, New Jersey,1978.

....to the execution time of the parallel algorithm on a given number of concurrent processors. The execution time corresponding to a parallel execution is the longest of the CPU times required by any of the concurrent processors. A general, commonsense definition of performance is presented in [6] wherein he defines performance as an indication of how well a system, already assumed to be correct, works. In general, the performance metric for parallel systems must reflect the program, the machine, and the implementation strategies for it depends on each one of them. However, while the ....

D. Ferrari. Computer Systems Performance Evaluation. Prentice-Hall Inc., Englewood Clis, New Jersey, 1978.

....analysis has formed the basis for the design of the solver and can help specification writers to understand how to set up experiments to obtain useful results from the solver. Many textbooks on performance include discussions of performance characterization or linear regression, for example, [10, 11, 18, 23]. 1.3.3 Complementary Performance Work PSpec is an approach to performance testing and debugging based on automatically checking expectations captured in performance assertions. Other work in performance testing and debugging has taken different approaches. Profiling tools, such as gprof [14] ....

....are relevant for a program, and what assertions can be made about those metrics. For guidance in this step, the discussion in Chapter 2 might be helpful in providing ideas about useful metrics, or the many books that have been written about performance evaluation and measurement might help (e.g. [10, 11, 18]) Obvious metrics to consider are response time, throughput, resource utilization, and workload properties (particularly where assumptions have been made in the implementation) Also, analysis, simulation, and measurement tools that provide insight into the performance of a program and help a ....

Domenico Ferrari. Computer Systems Performance Evaluation. Prentice-Hall, Inc., Englewood Cliffs, New Jersey 07632, 1978.

....information transparent, performance tuning is extremely difficult. P 3 T at compile time computes a set of performance parameters each of which reflects a different performance aspect. In the following all P 3 T performance parameters are described. 4. 1 Work Distribution It is well known [8, 6, 44, 42, 30, 40, 13, 41, 34, 25] that the work distribution has a strong influence on the cost performance ratio of a parallel system. An uneven work distribution may lead to a significant reduction in a program s performance. Therefore, providing both programmer and parallelizing compiler with a work distribution parameter for ....

D. Ferrari. Computer Systems Performance Evaluation. Prentice Hall, 1978.

....to eliminate all the negligible parts and include only those that best describe the real workload. In modeling workload, representativeness is one of the main characteristics that should be considered. The representativeness of a workload model can be determined by an equivalence relationship [Ferrari78] Given a system S, a set of performance indices L can be obtained from a certain workload W. Formally, a function f s can be defined for S, so that L f W s = If W r is a real workload and L r is the corresponding performance indices, then we have L f W r s r = Similarly, we have L f ....

Ferrari, D., Computer Systems Performance Evaluation, Prentice-Hall, 1978.

....The key functionality of P 3 T is devoted to compute a set of performance parameters at compile time: ffl Work Distribution The work distribution parameter describes how well the computations of a program are distributed over the set of available processors. As shown by numerous researchers [15, 13, 80, 68, 49, 62, 22, 67, 53, 34], work distribution has a strong influence on the cost performance ratio of a multiprocessor system. An uneven work distribution CHAPTER 3. P 3 T 25 may lead to a significant reduction in a program s performance. Therefore, providing both programmer and compiler with a work distribution ....

D. Ferrari. Computer Systems Performance Evaluation. Prentice Hall, 1978.

....speedup. In the following all activities related to modeling the real load are referred to as describing the load. ii) Specification When constructing artificial executable workload models, it is necessary to specify the load, i.e. the type of behavior, that the artificial workload has to mimic [6]. This specification can either be used to select several programs and benchmarks for a workload mix, or to parametrize a synthetic workload generator. Workload modeling of artificial executable test workloads will be called workload model specification. iii) Representation In a non executable ....

D. Ferrari. Computer Systems Performance Evaluation. Prentice-Hall, Englewood Cliffs, New Jersey, 1978.

....of the components. In Section 3 workload modeling is discussed in the context of a performance evaluation cycle. 2 Definition of Workload Traditionally, the workload is defined as the set of all inputs (processing requests) from the users to the processing system during a specified time interval [5]. Depending on the objective of the performance evaluation study, the workload may include programs, commands, and data submitted by the users, but also system programs (compilers, editors, requested by the users. The same components can be identified in a parallel system s workload. Because ....

D. Ferrari. Computer Systems Performance Evaluation. Prentice-Hall, Englewood Cliffs, New Jersey, 1978.

....deadline missed time reliability. Keywords: real time systems, performance evaluation, dynamic failure, static failure 1 Introduction Traditional analyses of computer systems are based on measures such as throughput, response time, utilization, availability, and reliability (for example, see [2]) These measures are not by themselves suitable to capture the essence of real time systems [5, 7] In real time systems, some of the measures of interest are dependability (encompassing reliability, availability, and safety) timeliness (the probability of missing meeting a task deadline or the ....

....a Markov model, 9] to describe a gracefully degrading multiprocessor soft real time system in which dynamic failures occur while hardware components are subject to failure. Queuing networks are by far one of the most appealing approaches to the study of computer systems performance evaluation [2]. This paper is organized as follows. Section 2 provides a brief survey of the related work. In Section 3 we describe our model and derive reliability and MTTF of the system. In Section 4 we introduce the new performance measures. In Section 5 we give some numerical examples and, finally, in ....

D. Ferrari. Computer Systems Performance Evaluation. Englewood Cliff, NJ Prentice-Hall, 1978.

....of the various notions of observables for a simple probabilistic program, and we show how they can be used for expressing the long run average running time of the program. We point out that, whereas for probabilistic programs the average has to be meant over an in nite period of time [13], for deterministic programs it is calculated over the (possibly in nite) di erent inputs. We also give an example for this latter case, which shows some similarities with the average case asymptotic analysis of algorithms. Finally, we suggest a possible application of the augmented operational ....

D. Ferrari. Computer Systems Performance Evaluation. Prentice-Hall, 1978.

..... He generates a 3 dimensional space to compare several architectures by local distances in space. This approach seems to be auspicious, but does not allow for a predictable architecture modeling. A predictable method to visualize the balance of a system are the kiviat graphs by Ferrari [Fer78] But this method is just supportive for modeling a system. 11 Conclusion There are many approaches dealing with the attempt of classifying computer systems. But most of them are merely sufficient even to describe special architectures. The major problem is, that at least, we have to combine ....

D. Ferrari. Computer Systems Performance Evaluation. Prentice Hall Inc., Englewood Cliff, New Jersey, U. S.A, 1978.

....information transparent, performance tuning is extremely difficult. P 3 T at compile time computes a set of performance parameters each of which reflects a different performance aspect. In the following all P 3 T performance parameters are described. 4. 1 Work Distribution It is well known [7, 5, 37, 35, 27, 33, 11, 34, 30, 23] that the work distribution has a strong influence on the cost performance ratio of a parallel system. An uneven work distribution may lead to a significant reduction in a program s performance. Therefore, providing both programmer and parallelizing compiler with a work distribution parameter for ....

D. Ferrari. Computer Systems Performance Evaluation. Prentice Hall, 1978.

....mechanisms for detecting deadlocks. Aborted jobs are not scheduled immediately but they are delayed for a number of rounds before restart. The execution time of all three simulators for a complete workload is about one day on a DECstation 5000 200. We also ensure through the method of batch means[9] that our simulations reach stability (confidence of more than 96 ) 5 Simulation Results In this section, we discuss performance metrics and describe some of the experiments conducted using the three simulators. System and data related parameter values appear in tables 1 to 4. 5.1 ....

D. Ferrari. Computer Systems Performance Evaluation. Prentice--Hall, Eaglewood Cliffs, NJ, 1978.

No context found.

D. Ferrari. Computer Systems Performance Evaluation. Prentice Hall, 1978.

No context found.

D. Ferrari. Computer Systems Performance Evaluation. Prentice Hall, 1978.

No context found.

D. Ferrari. Computer Systems Performance Evaluation. Prentice-Hall, Englewood Cliffs, New Jersey, 1978.

No context found.

D. Ferrari. Computer Systems Performance Evaluation. Prentice-Hall, Englewood Cliffs, New Jersey, 1978.

No context found.

D. Ferrari. Computer Systems Performance Evaluation. Prentice-Hall, Inc., 1978.

No context found.

D. Ferrari, ed., Computer Systems Performance Evaluation, Prentice-Hall, Englewood Cliffs, NJ, 1978.

No context found.

D. Ferrari, ed., Computer Systems Performance Evaluation, Prentice-Hall, Englewood Cliffs, NJ, 1978.

No context found.

D. Ferrari, Computer Systems Performance Evaluation, Prentice-Hall, 1978.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC