B. Carpenter, G. Zhang, and Y. Wen. NPAC PCRC runtime kernel definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. Up-to-date version maintained at http://www.npac.syr.edu/projects/pcrc/doc.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....other functionalities of collective communication operation are described. 4.1 Background A C library called Adlib [12] was completed in the Parallel Compiler Runtime Consortium (PCRC) 19] project. It was a high level runtime library designed to support translation of data parallel languages [16]. Initial emphasis was on High Performance Fortran (HPF) and two experimental HPF translators used the library to manage their communications [38, 58] It incorporated a built in representation of a distributed array, and a library of communication and arithmetic operations acting on these ....

Bryan Carpenter, Guansong Zhang, and Yuhong Wen. NPAC PCRC Runtime Kernel Definition. Technical Report CRPC-TR97726, Center for Research Parallel Computation, NPAC, Syracuse University, 1997. http://www.npac.syr.edu/projects/pcrc/kernel.html.

....characteristic and important collective operations, remap and writeHalo. Based on those operations, we describe detailed implementation issues of collective communication. Finally other functionalities of collective communication operation are described. 4. 1 Background A C library called Adlib [12] was completed in the Parallel Compiler Runtime Consortium (PCRC) 19] project. It was a high level runtime library designed to support translation of data parallel languages [16] Initial emphasis was on High Performance Fortran (HPF) and two experimental HPF translators used the library to ....

B. Carpenter, G. Zhang, and Y. Wen. NPAC PCRC runtime kernel definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. http://www.npac.syr.edu/projects/pcrc/kernel.html. 172

....transformational functions (including matrix multiplications and transposes, shift and spread operations, and all F90 scalar and array valued reduction functions) plus a subset of the HPF standard library. These functions operate uniformly on any array representable by the DAD described above [28]. The University of Florida subcontractors have developed an MPI version of their algorithms for combining scatter. This version is link compatible with the PCRC kernel developed at Syracuse. Syracuse has provided interface code so that these routines are now callable using the generic DAD array ....

....for parallel prefix and sorting operations (these are the only significant functions from the standard HPF library not implemented in the NPAC kernel library) 2. 4 Common Data Movement Routines Syracuse group has implemented an extensive set of collective data movement routines in their runtime [28, 83]. The communication operations supported include HPF F90 array intrinsic operations such as CSHIFT, the function pcrc write halo, which updates ghost areas of a distributed array, the function remap, which is equivalent to a Fortran 90 array assignment between a conforming pair of sections of two ....

[Article contains additional citation context not shown here]

Bryan Carpenter, Guansong Zhang, and Yuhong Wen. NPAC PCRC runtime kernel definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. Up-to-date version maintained at http://www.npac.syr.edu/projects/pcrc/docs.html.

No context found.

B. Carpenter, G. Zhang, and Y. Wen. NPAC PCRC runtime kernel definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. Up-to-date version maintained at http://www.npac.syr.edu/projects/pcrc/doc.

No context found.

B. Carpenter, G. Zhang, and Y. Wen. NPAC PCRC runtime kernel definition. Technical Report CRPCTR97726, Center for Research on Parallel Computation, 1997. Up-to-date version maintained at http://www.npac.syr.edu/projects/pcrc/doc.

....library oriented, SPMD style, omitting some basic assumptions of the HPF [12] model. To facilitate programming of massively parallel, distributed memory systems, it extends the Java language with some additional syntax and some pre defined classes for handling distributed arrays, and with Adlib [8], the run time communication library. HPJava supports a true multi dimensional array, which is a modest extension to the standard Java language, and which is a subset of our syntax for distributed arrays. HPJava introduces some new control constructs such as overall, at, and on statements. As ....

B. Carpenter, G. Zhang, and Y. Wen. NPAC PCRC Runtime Kernel Definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. Up-to-date version maintained at http://www.npac.syr.edu/projects/pcrc/doc.

....complex multidisciplinary problems. The Global Array (GA) library developed at the Pacific Northwest National Laboratory (PNNL) is one of the ACTS components. In ACTS the GA tool kit will be extended to provide support for higher dimensional arrays and additional distribution formats. The Adlib [4] parallel run time library has capabilities that can be used to implement these new features. Development of the Adlib library was completed at NPAC, Syracuse in the framework of the three year Parallel Compiler Runtime Consortium (PCRC) project. It incorporates experience gained over the years ....

....work on using Adlib based techniques in the implementation of GA. 2 Background 2.1 The Adlib library The Adlib library was completed in the Parallel Compiler Runtime Con sortium project [14] project. It is a high level runtime library designed to support translation of data parallel languages [4]. Initial emphasis was on High Performance Fortran (HPF) and two experimental HPF translators used the library to manage their communications [16, 10] Currently the library is being used in the HPspmd project at NPAC [3] It incorporates a built in representation of a distributed array, and a ....

B. Carpenter, G. Zhang, and Y. Wen. NPAC PCRC runtime kernel definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. http://www.npac.syr.edu/projects/pcrc/kernel.html.

....localized to some processors, and for writing a distributed form of the parallel loop. Crucially, it also supports binding from the extended languages to various communication and arithmetic libraries. These might involve simply new interfaces to some subset of PARTI [1] Global Arrays [10] Adlib [7], MPI [9] and so on. Providing the libraries for irregular communication may well be important. Evaluating HPspmd programming language model on large scale applications is also an important issue. 1 2 HPSpmd Programming Language Model 2.1 HPspmd Language Extensions In order to support a ....

Bryan Carpenter, Guansong Zhang, and Yuhong Wen. NPAC PCRC runtime kernel definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. Up-to-date version maintained at http://www.npac.syr.edu/projects/pcrc/doc.

.... used for several purposes, incuding implementation of the object based shared memory programming environment of Global Arrays [13] or ABC [8] and optimized implementation of collective operations on sections of multidimensional distributed arrrays, like those in the PCRC Adlib run time system [4] or the P [15] object oriented class library. In scientific computing, distributed array libraries communicate sections or fragments of sparse or dense multidimensional arrays. With remote copy APIs (application programming interfaces) supporting only contiguous blocks of data, array sections ....

....latency has increased by less than 0.5 S when comparing with the original implementation of GA. 3.2 Adlib The Adlib library was completed in the Parallel Compiler Runtime Consortium project [14] project. It is a high level runtime library designed to support translation of data parallel languages [4]. Initial emphasis was on High Performance Fortran, and two experimental HPF translators used the library to manage their communications [17, 11] Currently the library is being used in the HPspmd project at NPAC [3] It incorporates a built in representation of a distributed array, and a library ....

B. Carpenter, G. Zhang, and Y. Wen. NPAC PCRC runtime kernel definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. http://www.npac.syr.edu/projects/pcrc/kernel.html.

....complex multidisciplinary problems. The Global Array (GA) library developed at the Pacific Northwest National Laboratory (PNNL) is one of the ACTS components. In ACTS the GA tool kit will be extended to provide support for higher dimensional arrays and additional distribution formats. The Adlib [4] parallel run time library has capabilities that can be used to implement these new features. Development of the Adlib library was completed at NPAC, Syracuse in the framework of the three year Parallel Compiler Runtime Consortium (PCRC) project. It incorporates experience gained over the years in ....

....work on using Adlib based techniques in the implementation of GA. 2 Background 2.1 The Adlib library The Adlib library was completed in the Parallel Compiler Runtime Consortium project [14] project. It is a high level runtime library designed to support translation of data parallel languages [4]. Initial emphasis was on High Performance Fortran (HPF) and two experimental HPF translators used the library to manage their communications [16, 10] Currently the library is being used in the HPspmd project at NPAC [3] It incorporates a built in representation of a distributed array, and a ....

B. Carpenter, G. Zhang, and Y. Wen. NPAC PCRC runtime kernel definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. http://www.npac.syr.edu/projects/pcrc/kernel.html.

....(PCRC) project [14] PCRC was a DARPA supported project involving Rice, Maryland, Austin, Indiana, CSC, Rochester and Florida, with NPAC as prime contractor. Achievements included construction of an experimental HPF compilation system [39] delivery of the NPAC PCRC runtime kernel (Adlib) [11] and early work on the design and implementation of HPJava [10] 2 Objective and expected significance Our system aims to support a programming model that is a flexible hybrid of the data parallel, language oriented, HPF style, and the established and popular, library oriented, SPMD style. We ....

....pass in matrix with a distribution format unsupported by the ScaLAPACK routines. The interface code between HPspmd and ScaLAPACK (which converts between array descriptors) must either flag a runtime error in this case, or remap the argument array (using, for example, the remap primitive of Adlib [11]) In the second class we place libraries conceived primarily as underlying support for general parallel programs with regular distributed arrays. They emphasize high level communication primitives for particular styles of programming, rather than specific numerical algorithms. These libraries ....

[Article contains additional citation context not shown here]

Bryan Carpenter, Guansong Zhang, and Yuhong Wen. NPAC PCRC runtime kernel definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. Up-to-date version maintained at http://www.npac.syr.edu/projects/pcrc/doc.

....extensions that preserve many of the attractive features of HPF, while providing the programmer with direct access to runtime libraries such as the NPAC PCRC kernel. 1 Introduction As part of the PCRC [10] project we completed development of a high level runtime library for dataparallel languages [4]. The motivating goal was to simplify translation of High Performance Fortran (HPF) 6] by providing a coherent interface to the distributed array descriptors and collective operations needed for straightforward, efficient translation of parallel constructs like FORALL and array assignments. This ....

....overheads of the loop. Since distributed arrays are usually large, the loop ranges are typically large, and the startup overheads (including all the run time calls associated with address translation) can be amortized. This approach to translation of parallel loops is discussed in detail in [4]. Note that if array accesses are genuinely irregular, the necessary subscripting cannot usually be directly expressed in our language, because subscripts cannot be computed randomly in parallel loops without violating the fundamental SPMD restriction that all accesses be local. This is not ....

B. Carpenter, G. Zhang, and Y. Wen. NPAC PCRC runtime kernel definition. Technical Report CRPCTR97726, Center for Research on Parallel Computation, 1997. Up-to-date version maintained at http://www.npac.syr.edu/projects/pcrc/doc.

....the PNNL Global Array Toolset [8] in this connection. 3 Translation scheme The initial HPJava compiler is implemented as a source to source translator converting an HPJava program to a Java node program, with calls to runtime functions. The runtime system is built on the NPAC PCRC runtime library [3], which has a kernel implemented in C and a Java interface implemented in Java and C . 3.1 Java packages for HPspmd programming The current runtime interface for HPJava is called adJava. It consists of two Java packages. The first is the HPspmd runtime proper. It includes the classes needed to ....

....address of the first array element, then write the other addresses as a linear expression in the loop variable and this initial value. The code above can be rewritten in the form given in figure 6. This optimization again exposes various low level functions in the runtime we omit details (see [3]) The effect is to compute the parameters of the linear expressions for the local address offsets. This allows inlining of the element calls. In this case the resulting expressions are linear in the induction variables of the for loops. If necessary the multiplications can be eliminated by ....

Bryan Carpenter, Guansong Zhang, and Yuhong Wen. NPAC PCRC runtime kernel definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. Up-to-date version maintained at http://www.npac.syr.edu/projects/pcrc/docs.html.

.... care, it is possible to make direct calls to, say, MPI from within the data parallel program (in [2] we suggest a concrete Java binding for MPI) The language extensions described were devised partly to provide a convenient interface to a distributed array library developed in the PCRC project [5, 4]. Hence most of the run time technology needed to implement the language is available off the shelf . The existing library includes the run time descriptor for distributed arrays and a comprehensive array communication library. The HPJava compiler itself is being implemented initially as a ....

....overheads of the loop. Since distributed arrays are usually large, the loop ranges are typically large, and the startup overheads (including all the run time calls associated with address translation) can be amortized. This approach to translation of parallel loops is discussed in detail in [4]. Note that if array accesses are genuinely irregular, the necessary subscripting cannot usually be directly expressed in our language, because subscripts cannot be computed randomly in parallel loops without violating the fundamental SPMD restriction that all accesses be local. This is not ....

Bryan Carpenter, Guansong Zhang, and Yuhong Wen. NPAC PCRC runtime kernel definition. Technical Report CRPC-TR97726, Center for Research on Parallel Computation, 1997. Up-to-date version maintained at http://www.npac.syr.edu/projects/pcrc/doc.

....where the meaning is insensitive to the detailed mapping of elements. Lower level styles of programming are also possible. Our compiler will be implemented as a translator to ordinary Java with calls to a suitable run time library. At the time of writing the underlying library is already available [5], and the Java interface needed by the translator is under development. The translator itself is being implemented in a compiler construction framework developed in the PCRC project [6, 12] 2 Multidimensional arrays First we describe a modest extension to Java that adds a class of true ....

....that coexists in the language with ordinary Java arrays. There are good technical reasons for keeping the two kinds of array separate 1 . The type signatures and constructors of the multidimensional array use double brackets to distinguish them from ordinary arrays: int [ a = new int [[5, 5]] float [ b = new float [ 10, n, 20] int [ c = new int [ 100] a, b and c are respectively 2 , 3 and one dimensional arrays. Of course c is very similar in structure to the standard array d, created by int [ d = new int [100] c and d are not identical, though 2 . Access to ....

[Article contains additional citation context not shown here]

Bryan Carpenter, Guansong Zhang, and Yuhong Wen. NPAC PCRC Runtime Kernel Definition, 1997. In preparation. For current draft, see http://www.npac.syr.edu/projects/pcrc/July97/doc.html.

No context found.

B. Carpenter, G. Zhang, and Y. Wen. NPAC PCRC runtime kernel definition. Technical Report CRPCTR97726, Center for Research on Parallel Computation, 1997. Up-to-date version maintained at http://grids .ucs.indiana.edu/ptliupages/projects/HPJava/pcrc/docs .html.

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