Functional programming languages are well suited to the expression and automatic mapping of parallel computations. The Paradigm compiler is being developed to automatically analyze and map a class of Sisal programs am...
详细信息
Functional programming languages are well suited to the expression and automatic mapping of parallel computations. The Paradigm compiler is being developed to automatically analyze and map a class of Sisal programs amenable to static analysis for execution by distributed computer systems. A program description tree is constructed to represent the source program. In this form, the program is transformed into a hierarchy of acyclic interconnections of program modules (code blocks) of two basic types: array generators and stream producers. The program description tree is then used to guide decisions about allocation of processing elements to code blocks, and to construct of code for a target multiprocessor. We discuss the problem of finding an optimal allocation (mapping), and illustrate the methodology using a practical signal processing example.
This paper describes the integration of nested data parallelism into imperative languages using the example of C. Unlike flat data parallelism, nested data parallelism directly provides means for handling irregular da...
详细信息
This paper describes the integration of nested data parallelism into imperative languages using the example of C. Unlike flat data parallelism, nested data parallelism directly provides means for handling irregular data structures and certain forms of control parallelism, such as divide-and-conquer algorithms, thus enabling the programmer to express such algorithms far more naturally. Existing work deals with nested data parallelism in a functional environment, which does help avoid a set of problems, but makes efficient implementations more complicated. Moreover, functional languages are not readily accepted by programmers used to languages, such as Fortran and C, which are currently predominant in programmingparallel machines. In this paper, we introduce the imperative data-parallel language V and give an overview of its implementation.
In this paper we show that, under different circumstances, data scheduling and loop scheduling are both useful models for parallel programs executing on shared virtual memory (SVM) systems. We therefore propose a unif...
详细信息
In this paper we show that, under different circumstances, data scheduling and loop scheduling are both useful models for parallel programs executing on shared virtual memory (SVM) systems. We therefore propose a unified programming model that permits both types of scheduling. We show that, given affine array references, a program segment which is parallel under loop scheduling can always be transformed to make it parallel under data scheduling and vice-versa, and hence that the two types of scheduling are equally powerful at exploiting parallelism. We review existing Fortran dialects for SVM and propose compiler directives that allow program segments to be data scheduled.
Generalized Stochastic Petri Nets (GSPN) have gained a wide acceptance as a modeling tool for the performance analysis of concurrent systems. However, the applicability of this methodology is severely limited by the p...
详细信息
ISBN:
(纸本)3540600299
Generalized Stochastic Petri Nets (GSPN) have gained a wide acceptance as a modeling tool for the performance analysis of concurrent systems. However, the applicability of this methodology is severely limited by the potential state space explosion phenomenon. In this paper we describe massivelyparallel approaches to the most computing-intensive part of the solution of GSPN models: the state space construction. The effectiveness of these parallel approaches stays, for every GSPN, in their ability to deal with very large reachability spaces in reasonable time. Both the SIMD and the MIMD programmingmodels are considered, and examples are given using recent massivelyparallel processing architectures (CM-5, T3D).
This paper presents analytical models for evaluating intermediate tests for yield enhancement and quality assurance of systems manufactured using fault-tolerant multichip modules (MCM's) for massivelyparallel com...
详细信息
This paper presents analytical models for evaluating intermediate tests for yield enhancement and quality assurance of systems manufactured using fault-tolerant multichip modules (MCM's) for massivelyparallel computing (MPG), In the proposed approaches, we employ both a novel Markov model and a so called working-test-set to compute the yield, Unlike a previous method [1] which utilizes a binomial distribution, our scheme can employ intermediate tests to meet MCM quality requirements effectively. Several strategies for appropriately testing fault-tolerant MCM's have been proposed, but little analytical evaluation has been performed, In this paper, it is shown that an efficient test strategy with a modest level of redundancy may exist to achieve virtually 100% first-pass MCM yield for a particular system, We note that a yield-analysis model employing the LRTWS (Least Recently Tested in WS) test strategy proposed in this paper may provide a very good figure of merit due to its cost, delivery, number of tests and reliability benefits for current technology, Extensive parametric results for the analysis are provided to show that our approach can be applied to calculate the overall yield for fault-tolerant MCM's more accurately and efficiently, thereby improving upon the reliability and quality of the entire system.
Today’s massivelyparallel machines are typically message passing systems consisting of hundreds or thousands of processors. Implementing parallel applications efficiently in this environment is a challenging task. T...
详细信息
The increasing use of massivelyparallel supercomputers to solve large-scale scientific problems has generated a need for tools that can predict scalability trends of applications written for these machines. Much work...
详细信息
The increasing use of massivelyparallel supercomputers to solve large-scale scientific problems has generated a need for tools that can predict scalability trends of applications written for these machines. Much work has been done to create simple models that represent important characteristics of parallel programs, such as latency, network contention, and communication volume. But many of these methods still require substantial manual effort to represent an application in the model's format. The MK toolkit described in this paper is the result of an on-going effort to automate the formation of analytic expressions of program execution time, with a minimum of programmer assistance. In this paper we demonstrate the feasibility of our approach, by extending previous work to detect and model communication patterns automatically, with and without overlapped computations. The predictions derived from these models agree, within reasonable limits, with execution times of programs measured on the Intel iPSC/860 and Paragon. Further, we demonstrate the use of MK in selecting optimal computational grain size and studying various scalability metrics.
作者:
Cosnard, M.LIP-CNRS
Ecole Normale Supérieure de Lyon 46 Alée d'Italie Lyon69364 France
We compare various models of parallel machines and show that they can be classified in two classes: algorithm oriented or execution oriented. None of them are really satisfying from the user's point of view. Hence...
详细信息
This paper describes a research proposal related to the design of parallel software for image processing. The proposal focuses on the design of a tool, which generates the best implementation of a given application ta...
详细信息
暂无评论