Automatic Support for Data Distribution

Chapman, Barbara; Herbeck, Heinz M.; Zima, Hans P.

doi:10.1109/dmcc.1991.633085

Cited by 21 publications

(6 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Consequently, those machines will also profit from this optimization (perhaps with different implementation of the runtime communication routines). [8,48] currently only supports arbitrary rectilinear block distributions. The ADAPT system [35], developed at University of Southampton, compiles Fortran 90 for execution on MIMD distributed-memory machines.…”

Section: Discussionmentioning

confidence: 99%

An Algebraic Machinery for Optimizing Data Motion for HPF

Chen

1997

Scientific Programming

View full text Add to dashboard Cite

This paper describes a general compiler optimization technique that reduces communication over-head for FORTRAN-90 (and High Performance FORTRAN) implementations on massively parallel machines. The main sources of communication, or data motion, for the parallel implementation of a FORTRAN-90 program are array assignments, array operators (e.g., CSHIFT, TRANSPOSE, etc.), and array parameters passing to and from subroutines. Coupled with the variety of ways arrays can be distributed, a FORTRAN-90 implementor faces a rich space of posibilities by which data motion can be organized. We propose a unified framework for optimizing intra-and inter-procedural data motion. The central idea of this framework is algebraic analysis of data motion. We give an algebraic representation for each HPF's array intrinsics and data distribution specifications. An array reference extracted from the source FORTRAN-90 program, given a particular data distribution specification, is represented as a communication expression, which in turn can be simplified according to a communication algebra. Fast communication is uncovered by pattern matching with a set of communication idioms. Experimental results on the Connection Machine CM-5 demonstrating the effectiveness of this approach are reported.

show abstract

Section: Discussionmentioning

confidence: 99%

An Algebraic Machinery for Optimizing Data Motion for HPF

Chen

1997

Scientific Programming

View full text Add to dashboard Cite

show abstract

“…Vienna Fortran and Vienna Fortran-90, based upon the parallelizing system SUPERB, extends Fortran and Fortran-90 by providing alignment and distribution specifications. The Vienna Fortran compiler [41,42] focuses on supporting arbitrary rectilinear block distributions and irregular data distribution through indirect arrays. The SHPF compilation system [43] supports a full set of HPF data distribution directives.…”

Section: Related Workmentioning

confidence: 99%

CRAFT: a framework for F90/HPF compiler optimizations

Chen

Cowie

1999

Concurrency: Pract. Exper.

View full text Add to dashboard Cite

In this paper, we give an overview of the results of the CRAFT optimising compiler project (Fortran 90/HPF subset compilers). We start by describing the theoretical framework within which we designed program transformations for the optimization of inter‐ and intra‐procedural data motion, as well as the optimizations for parallel loops; we then describe the implementation of the CRAFT compilers for Thinking Machines' CM‐2 and CM‐5. We report results from experiments on the Connection Machine CM‐5, the IBM SP‐2 and a network of UltraSparc workstations. The results demonstrate that these optimizations can achieve significant object code performance improvement. Copyright © 1999 John Wiley & Sons, Ltd.

show abstract

“…Instead of nding the set of all cycles, we restrict our considerations to the set of fundamental cycles 3 in P that bear a con ict. Nothing is lost by this restriction since all cycles in P can be constructed by combinations of fundamental cycles.…”

Section: Fundamental Cyclesmentioning

confidence: 99%

“…Some of those require that data is mapped onto the topology by hand 24,32], others are user guided and o er sets of directives for the compiler, abstract topologies (so-called templates), or interactive or knowledge-based environments that help determine the alignment of array dimensions and mapping functions 3,15,16].…”

Section: Introductionmentioning

confidence: 99%

Automatic alignment of array data and processes to reduce communication time on DMPPs

Philippsen

1995

SIGPLAN Not.

View full text Add to dashboard Cite

This paper investigates the problem of aligning data and processes in a distributed-memory implementation. We present complete algorithms for compile-time analysis, the necessary program restructuring, and subsequent code-generation, and discuss their complexity. W e nally evaluate the practical usefulness by quantitative experimentation.The technique presented analyzes complete programs, including branches, loops, and nested parallelism. Alignment is determined with respect to o set, stride, and general axis relations. Both placement of data and processes are computed in a unifying framework based on an extended preference graph and its analysis. Furthermore, dynamic redistribution and replication are considered in the same technique.The experimental results are very encouraging. The optimization algorithms implemented in the Modula-2* compiler improved the execution times of the programs by o ver 40% on a MasPar MP-1 with 16384 processors.

show abstract

Automatic Support for Data Distribution

Cited by 21 publications

References 3 publications

An Algebraic Machinery for Optimizing Data Motion for HPF

An Algebraic Machinery for Optimizing Data Motion for HPF

CRAFT: a framework for F90/HPF compiler optimizations

Automatic alignment of array data and processes to reduce communication time on DMPPs

Contact Info

Product

Resources

About