Ian Foster, Ming Xu, Bhaven Avalani, and Alok Choudhary. A Compilation System That Integrates High Performance Fortran and Fortran M. In Proc. 1994 Scalable High Performance Computing Conf., page July. IEEE Computer Science Press, 94.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....this type of parallelism. These examples are by no means exhaustive merely representative. One difficulty with identifying applications is that the tools available influence how, and which, applications are coded. Recent activity in the field of combined task and data parallel language design [3, 5, 7, 22] is further evidence of the importance of the problem. We have designed a set of data parallel extensions to the Mentat Programming Language (MPL) which allow us to integrate task parallelism, data parallelism, and nested task and data parallelism within a single language on top of a single run ....

....operations. This characteristic requires that the body of the function be written without reference to any specific element. In scale( the member variable pixel is assigned its value multiplied by the argument value. No reference is made by the programmer to a particular element such as object[5][10] pixel. The compiler will use the information contained in the annotations and the invocation of the method to index into the data set properly. Mechanism is provided to allow the programmer to reference other elements of the data set in a relative fashion. The starting point is the element ....

[Article contains additional citation context not shown here]

I. Foster, M. Xu, B. Avalani, A. Choudhary, "A Compilation System that Integrates High Performance Fortran and Fortran M", Proccedings of the 1994 Scalable High Performance Computing Conference.

....and communication among tasks. Linda is described in detail below. Therefore, HPF s tasking features are mainly suitable for algorithms which do not require intertask communication, such as divide and conquer. Several proposals for more advanced tasking facilities in HPF have been made [33, 51, 74, 84, 96, 139, 157]. Inter process communication in HPF is due to non local data accesses. An HPF compiler is in duty of generating appropriate code, in most cases message passing code, that takes care of providing non local data in place. All communication is handled transparently to the user. Linda [3, 45] is a ....

I. Foster, B. Avalani, M. Xu, and A. Choudhary. A Compilation System That Integrates High Performance Fortran and Fortran M. In Proceedings 1994.

....models support the programmer in writing efficient programs without dealing too much with the underlying communication and coordination details of a specific parallel machine. Language approaches include Braid, Fortran M, Fx, Opus, and Orca, see [2] for an overview and comparison. Fortran M [8, 7] allows the creation of processes which can communicate with each other by predefined channels and which can be combined with HPF Fortran for a mixed task and data parallel execution. The organization of communication in Fortran M is different than within the Tlib library, since communication in ....

I. Foster, M. Xu, B. Avalani, and A. Choudhary. A Compilation System That Integrates High Performance Fortran and Fortran M. In Proceedings 1994.

....groups. Previous work on supporting multi level parallelism focused on providing some kind of coordination support to allow the interaction of a set of program modules (task parallelism) in the framework of HPF [11] parallel programs (data parallelism) for distributed memory architectures [5,6,8,26]. Recently, KAI (Kuck and Associates, Inc. has made proposals to OpenMP to support multi level parallelism through the WorkQueue mechanism [12] in which work can be created dynamically, even recursively, and put into queues. Within the WorkQueue model, nested queuing permits a hierarchy of ....

I. Foster, B. Avalani, A. Choudhary and M. Xu. A Compilation System that Integrates High Performance Fortran and Fortran M. Scalable High Performance Computing Conference, Knoxville (TN), May 1994.

....the interaction of a set of program modules in the framework of data parallel programs for distributed memory architectures. For example, 3] proposed a library based approach that provides a set of functions for coupling multiple HPF tasks to form task parallel computations. Other alternatives [2, 6, 10] proposed a small set of Fortran directives to integrate task and data parallelism also in an HPF framework. Our group has proposed extensions to OpenMP in order to allow an efficient exploitation of nested parallelism. The extensions offer the concept of thread groups [5] This concept is similar ....

....numsucc and matrix listsucc) Different heuristics have been proposed in the literature for doing this assignment [9, 12, 10] For instance, assume that the following list of successors is specified (corresponding to the graph shown in Figure 2. a: numsucc[8] 2, 1, 0, 2, 1, 0, 1, 0 listsucc[8,2] = 2, 4, 3, 0, 0, 0, 5, 7, 6, 0, 0, 0, 8, 0, 0, 0 In this case, for instance block number 4 has two successors (as indicated by numsucc[4] 2) blocks 5 and 7 (as indicated by the two valid entries in listsucc[4] f5, 7g) A possible allocation strategy could end up clustering groups as shown in ....

I. Foster, B. Avalani, A. Choudhary and M. Xu. A Compilation System that Integrates High Performance Fortran and Fortran M. In Scalable High Performance Computing Conference, Knoxville (TN), May 1994.

....programs for distributed memory architectures. Some of them combine the use 1 of two programming models and interfaces. For example, 3] proposes a library based approach that provides a set of functions for coupling multiple HPF tasks to form task parallel computations. Other alternatives [2, 5, 16] proposed a small set of Fortran directives to integrate task and data parallelism parallelism also in an HPF framework. The Illinois Intel Multithreading library [4] targets shared memory systems. It also supports multiple levels of general (unstructured) parallelism. Application tasks are ....

I. Foster, B. Avalani, A. Choudhary and M. Xu. A Compilation System that Integrates High Performance Fortran and Fortran M. In Scalable High Performance Computing Conference, Knoxville (TN), May 1994.

....and communication among autonomous tasks. Task parallelism provides no explicit support for data decomposition. Task parallelism is ideally suited for computations integrating various heterogenous operations, such as a multidisciplinary simulation coordinating various independent components [76]. However, it is inappropriate for the coarse grain scientific applications addressed by LPARX, which are more naturally expressed in a coarsegrain data parallel fashion (see Section 2.2.3) 41 Split C Split C [57] is a parallel extension to C for distributed memory multiprocessors. Split C gets ....

.... how should the programmer in a task parallel language specify a call to a data parallel routine How should data be transferred across the call interface HPF [83] defines a rudimentary external procedure interface, and others have investigated calling HPF from pC [144] and also from Fortran M [76]. However, generally applicable mechanisms are currently unknown. 6.3 The Scientific Computing Community Build it, and they will come: Field of Dreams The goal of our research has been the development of software tools to enable computational scientists to explore new approaches in ....

I. Foster, M. Xu, B. Avalani, and A. Choudhary, A compilation system that integrates High Performance Fortran and Fortran M, in Proceedings of the 1994 Scalable High Performance Computing Conference, May 1994.

....further models. Related work on performance prediction includes the studies in the context of parallelizing compilers [9, 21] as well as other prediction models [7, 5] Several research groups working on parallelizing compilers have included support to combine task and data parallelism. Fortran M [16, 11] allows the creation of processes which can communicate with each other by predefined channels and which can be combined with HPF Fortran for a mixed task and data parallel execution. In contrast to the Fortran M approach, the TwoL approach is more directed towards the combination of fine and ....

I.Foster, M.Xu, B.Avalani, and A.Choudhary. A Compilation System That Integrates High Performance Fortran and Fortran M. In Proccedings 1994 Scalable High Performance Computing Conference. IEEE Computer Society Press, 1994.

....costs on distributed memory machines (DMMs) It is often difficult to choose the most efficient implementation. Since much better performance can be achieved by exploiting both forms of parallelism, a variety of approaches have been developed to combine them. These include the Fx [21] Fortran M [9], and PARADIGM [2] projects. Mainly, these are projects in the context of parallelizing compilers, which transform sequential programs into parallel output programs. A detailed discussion is given later. author supported by DFG In this article, we present the integrated framework TwoL (Two Level ....

....compilers [7, 12] as well as other prediction models [6] In the following, we compare the TwoL approach to similar approaches in the area of parallelizing compilers. Several research groups working on parallelizing compilers have included support to combine task and data parallelism. Fortran M [9, 8] allows the creation of processes which can communicate with each other by predefined channels and which can be combined with HPF Fortran for a mixed task and data parallel execution. In contrast to the Fortran M approach, the TwoL approach is more directed towards the combination of fine and ....

I. Foster, M. Xu, B. Avalani, and A. Choudhary. A Compilation System That Integrates High Performance Fortran and Fortran M. In Proccedings 1994 Scalable High Performance Computing Conference. IEEE Computer Society Press, 1994.

.... parallel languages [4, 18, 9] parallelizing compilers [3] and runtime prediction models [12, 2] A more detailed discussion of this work can be found in [26] There is also some relationship between our work and the work on combining task and data parallelism as it is considered in Fortran M [15], Fx [29] and the Paradigm compiler [5] In Fx, the programmer has the possibility to express task parallelism by parallel sections containing subroutine calls. Each call is followed by input and output directives identifying the input and output parameters of the subroutine. Each subroutine call ....

I. Foster, M. Xu, B. Avalani, and A. Choudhary. A Compilation System That Integrates High Performance Fortran and Fortran M. In Proccedings 1994 Scalable High Performance Computing Conference. IEEE Computer Society Press, 1994.

....modern parallel machines support MIMD (Multiple Instruction Multiple Data) execution, and therefore combined data and task (or function) parallel computing. Compiler and runtime support for task and data parallel computing is an active area of research, and several solutions have been proposed [4, 5, 9, 10, 19, 21]. Recent research has also examined the benefits of mixed task and data parallel programming [3, 7, 18] This paper specifically addresses the mapping of applications composed of a linear chain of data parallel tasks that act on a stream of input data sets. In this model, each task repeatedly ....

FOSTER, I., AVALANI, B., CHOUDHARY, A., AND XU, M. A compilation system that integrates High Performance Fortran and Fortran M. In Proceeding of 1994 Scalable High Performance Computing Conference (Knoxville, TN, October 1994), pp. 293--300.

....computation loads in [25, 29] With the standardization of High Performance Fortran [12] several commercial compilers for the language are also available. Support for task parallelism in Fx is discussed in [28] and the subject has also been addressed by several other research efforts, including [3, 7, 11]. The Airshed model for pollution modeling was developed by McRae et al. 17] and a parallel implementation, based on PVM, was developed by Kumar et al. 16] A parallel implementation of the original uniform grid CIT model is described by Dabdub et al. 5] To our knowledge, this is the ....

FOSTER, I., AVALANI, B., CHOUDHARY, A., ANDXU, M. A compilation system that integrates High Performance Fortran and Fortran M. In Proceeding of 1994 Scalable High Performance ComputingConference (Knoxville, TN, October 1994), pp. 293--300.

....similar to HPF and functional parallelism is specified using constructs called Parallel Sections. The FORTRAN M language inherently provides constructs for specifying functional parallelism [18] recent work proposes to integrate the language with HPF in order to specify data parallelism [34]. The work in [19] has proposed extensions to FORTRAN 90 for specifying functional parallelism. In the next section, we provide a brief overview of the theory of convex and posynomial functions which we use in developing our allocation algorithm. In the following sections we discuss our allocation ....

I. Foster, M. Xu, B. Avalani, and A. Choudhary, "A Compilation System That Integrates High Performance FORTRAN and FORTRAN-M," in the proceedings of the Scalable High Performance Computing Conference, (Knoxville, TN), pp. 293--300, May 1994.

....of processors on a parallel machine. Most modern parallel machines support MIMD execution, and therefore combined data and task (or function) parallel computing. Compiler and runtime support for task and data parallel computing is an active area of research and several solutions have been proposed [2, 3, 7, 8, 13]. Recent research has also examined the tradeoffs involved in mapping task and data parallel programs [5, 12] In this paper we address the problem of optimizing the throughputof task and data parallel programs. We address applications composed of a linear chain of data parallel tasks that act on ....

FOSTER, I., AVALANI, B., CHOUDHARY, A., AND XU, M. A compilation system that integrates High Performance Fortran and Fortran M. In Proceeding of 1994 Scalable High PerformanceComputingConference (Knoxville, TN, October 1994), pp. 293--300.

....We have started an initial implementation of the above design and will experiment with the various choices in order to characterize the overheads involved with each of them. 4 Related Research Other language projects that focus on the integration of task and data parallelism include FortranM [12, 13], DPC [30] and FX [31, 32] Fortran M and DPC support mechanisms for establishing message pathways between tasks, and all communication between the tasks is transmitted via these pathways. Fortran M uses the concept of channels (similar to a Unix pipe [1] for establishing its pathways, and DPC ....

I. Foster, M. Xu, B. Avalani, and A. Choudhary. A compilation system that integrates High Performance Fortran and Fortran M. In Scalable High Performance Computing Conference, May 1994.

....parallelism in a language independent and environment independent manner. Programmers can use our archetypes with languages and tools with which they are familiar. Many of the ideas on compositionality described here are based on ideas in Fortran M [12] The integration of Fortran M and Fortran D [7, 11] is particularly relevant. Likewise, the integration of pC [4] with CC [8, 6] provides an object oriented language that allows for integration of task and data parallelism. Merlin [9] also takes an abstraction based approach to integrating task and data parallelism, but it uses monitors rather ....

I. T. Foster, B. Avalani, A. Choudhary, and M. Xu. A compilation system that integrates High Performance Fortran and Fortran M. In Proceedings: 1994 Scalable High Performance Computing Conference. IEEE, 1994. To appear.

....prediction techniques [9, 17, 19] and parallelizing compilers. In the following, we compare our approach to similar approaches in the area of parallelizing compilers. Several research groups working on parallelizing compilers have included support to combine task and data parallelism. Fortran M [11, 12] allows the creation of processes which can communicate with each other by predefined channels and which can be combined with HPF for a mixed task and data parallel execution. In contrast to the Fortran M approach, our approach is more directed towards the combination of fine and medium grain ....

I. Foster, M. Xu, B. Avalani, and A. Choudhary. A Compilation System That Integrates High Performance Fortran and Fortran M. In Proccedings 1994 Scalable High Performance Computing Conference. IEEE Computer Society Press, 1994.

....loosely synchronous parallel execution model. Many problems requiring high performance implementations can be expressed succinctly in HPF. However, HPF does not adequately address task parallelism or heterogeneous computing. Examples of applications that are not easily expressed using HPF alone [6, 14] include multidisciplinary applications where different modules represent distinct scientific disciplines, programs that interact with user interface devices, applications involving irregularly structured data such as multiblock codes, and image processing applications in which pipeline structures ....

....Two principal approaches have been investigated. Compiler based approaches seek to identify task parallel structures automatically, within data parallel specifications [11, 14, 21] while language based approaches provide new language constructs for specifying task parallelism explicitly [3, 6, 19, 24]. Both approaches have shown promise in certain application areas, but each also has disadvantages. Compiler based approaches complicate compiler development and performance tuning, and language based approaches also introduce the need to standardize new language features. In this paper, we ....

[Article contains additional citation context not shown here]

I. Foster, B. Avalani, A. Choudhary, and M. Xu. A compilation system that integrates High Performance Fortran and Fortran M. In Proceedings of the 1994 Scalable High Performance Computing Conference, pages 293--300. IEEE Computer Society Press, 1994.

....FM is designed to support both the modular construction of large parallel programs and the development of libraries implementing other programming paradigms. For example, in a joint project with Syracuse, such a library has been used to integrate HPF programs into a task parallel framework [12]. FM programs can dynamically create and destroy processes, single reader singlewriter channels, and multiple writer, single reader mergers. Processes can encapsulate state (common data) and communicate by sending and receiving messages on channels and mergers; references to channels, called ....

I. Foster, B. Avalani, A. Choudhary, and M. Xu. A compilation system that integrates High Performance Fortran and Fortran M. In Proc. 1994 Scalable High Performance Computing Conf. IEEE, 1994. to appear.

No context found.

Ian Foster, Ming Xu, Bhaven Avalani, and Alok Choudhary. A Compilation System That Integrates High Performance Fortran and Fortran M. In Proc. 1994 Scalable High Performance Computing Conf., page July. IEEE Computer Science Press, 94.

No context found.

I. Foster, B. Avalani, A. Choudhary, and M. Xu, "A Compilation System That Integrates High Performance Fortran and Fortran M," in Proceedings of the Scalable High Performance Computing Conference, Knoxville, TN, May 1994, pp. 293--300.

No context found.

Foster, I., Xu, M., Avalani, B. and Choudhary, A. A Compilation System That Integrates High Performance Fortran and Fortran M. In

No context found.

Version 1.0. #34# I. Foster, M. Xu, B. Avalani, and A. Choudhary, #A Compilation System That Integrates High Performance FORTRAN and FORTRAN-M," in the proceedings of the Scalable High Performance Computing Conference, #Knoxville, TN#, pp. 293#300, May 1994.

No context found.

Foster, I., Xu, M., Avalani, B. and Choudhary, A. A Compilation System That Integrates High Performance Fortran and Fortran M. In 1994 Scalable High Performance Computing Conference, May 1994.

No context found.

Foster, I., Xu, M., Avalani, B. and Choudhary, A. A Compilation System That Integrates High Performance Fortran and Fortran M. In 1994 Scalable High Performance Computing Conference, May 1994.

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