High Performance Fortran Forum. Hpf language specication, version 2.0. Available from http://www.crpc.rice.edu/HPFF/versions/hpf2/les/hpfv20. ps.gz, January 1997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....iterations can be performed in parallel. We also extend the semantics of foreach to include the possibility of updates to reduction variables, as we explain later. We introduce a Java interface called Reducinterface. Any object of any class implementing this interface acts as a reduction variable [18]. The semantics of a reduction variable is analogous to that used in version 2.0 of High Performance Fortran (HPF 2) 18] and in HPC [5] A reduction variable has the property that it can only be updated inside a foreach loop by a series of operations that are associative and commutative. ....

....reduction variables, as we explain later. We introduce a Java interface called Reducinterface. Any object of any class implementing this interface acts as a reduction variable [18] The semantics of a reduction variable is analogous to that used in version 2. 0 of High Performance Fortran (HPF 2) [18] and in HPC [5] A reduction variable has the property that it can only be updated inside a foreach loop by a series of operations that are associative and commutative. Furthermore, the intermediate value of the reduction variable may not be used within the loop, except for self updates. 4.2 ....

[Article contains additional citation context not shown here]

High Performance Fortran Forum. Hpf language specication, version 2.0. Available from http://www.crpc.rice.edu/HPFF/versions/hpf2/les/hpfv20. ps.gz, January 1997.

....parallelism is generally easier to exploit due to the simpler computational model involved. In addition, data parallel languages are available which substantially help the programmer to develop data parallel programs. High Performance Fortran (HPF) is the most notable example of such languages [13]. Unfortunately, data parallelism alone does not allow many potentially parallel applications to be dealt with eciently. Task parallelism is in fact often needed to re ect the natural structure of the application algorithm or to eciently implement applications that exhibit only a limited amount of ....

....a single framework is the subject of many proposals since it allows the number of addressable applications to be considerably enlarged [11; 16; 10; 8; 6; 14] Also the latest version of the HPF standard, HPF 2. 0, provides extensions which allows restricted forms of task parallelism to be exploited [13]. Unfortunately, the extensions are not suitable for expressing complex interactions among asynchronous tasks as required by multidisciplinary applications. Moreover, communication among tasks can occur only implicitly at the subroutine boundaries, and non deterministic communication patterns ....

[Article contains additional citation context not shown here]

High Performance Fortran Forum. HPF Language Speci cation, Jan. 1997. Ver. 2.0.

....This work has been partially supported by HPPC SEA contract number EU1063. number of addressable applications to be considerably enlarged [11,22,10,8,6,15] Also the latest version of the HPF standard, HPF 2. 0, provides extensions which allows restricted forms of task parallelism to be exploited [16]. Unfortunately, the extensions are not suitable for expressing complex interactions among asynchronous tasks as required by multidisciplinary applications. Moreover, communication among tasks can occur only implicitly at the subroutine boundaries, and non deterministic communication patterns ....

High Performance Fortran Forum. HPF Language Specication, Jan. 1997. Ver. 2.0.

....parallelism is generally easier to exploit due to the simpler computational model involved. In addition, data parallel languages are available which substantially help the programmer to develop data parallel programs. High Performance Fortran (HPF) is the most notable example of such languages [14]. Unfortunately, data parallelism alone does not allow many potentially parallel applications to be dealt with eciently. Task parallelism is in fact often needed to re ect the natural structure of the application algorithm or to eciently implement applications that exhibit only a limited amount of ....

....to exchange simple and structured data through these channels. To obtain portability among di erent HPF compilation systems, COLT HPF exploits HPF standard features alone and does not rely on compiler speci c features. Its HPF interface consists of a library of HPF LOCAL EXTRINSIC subroutines [14]. This means that when a COLT HPF primitive is invoked by an HPF task, all the processors executing the task switch from the single thread of the control model supplied by HPF to a true SPMD style of execution. In other words, through a LOCAL routine programmers can get the SPMD program generated ....

High Performance Fortran Forum. HPF Language Specication, May 1993. Ver. 1.0.

....a single framework is the subject of many proposals since it allows the number of addressable applications to be considerably enlarged [6, 19, 5, 3, 1, 10] Also the latest version of the HPF standard, HPF 2. 0, proposes an unintuitive and in exible approach to express pipelining among HPF tasks [11]. A di erent approach regards the design of an HPF binding for a standard message passing library such as MPI [5] The integration of task and data parallelism has also been proposed within an object oriented coordination language such as Opus [3] By using a syntax similar to that of HPF, Opus ....

High Performance Fortran Forum. HPF Language Specication, Jan. 1997. Ver. 2.0.

....parallelism is generally easier to exploit due to the simpler computational model involved. In addition, data parallel languages are available which substantially help the programmer to develop dataparallel programs. High Performance Fortran (HPF) is the most notable example of such languages [9]. Unfortunately, data parallelism alone does not allow many potentially parallel applications to be dealt with e ciently. Task parallelism is in fact often needed to re ect the natural structure of the application algorithm or to e ciently implement applications that exhibit only a limited amount ....

....conclusions and outlines future work. 2 COLT HPF implementation To obtain portability among di erent HPF compilation systems, COLT HPF exploits HPF standard features alone and does not rely on compiler speci c features. Its HPF interface consists of a library of HPF LOCAL EXTRINSIC subroutines [9]. A previous version of COLT HPF [17] exploited MPI for both inter and intra task communications, and required slight modi cations to the HPF run time in order to avoid interferences between the two communication contexts. The need to modify the HPF run time meant that commercial black box HPF ....

High Performance Fortran Forum. HPF Language Specication, May 1993. Ver. 1.0.

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