Automatic decomposition in EPPP compiler
نویسندگان
چکیده
In this paper, we give an overview of our automatic computation and data decomposition scheme in the EPPP compiler 1 for distributed memory machines. Such automatic techniques can be also be used in compilers for programminglanguages such as HPF (High-Performance Fortran) in conjunction with the user-speciied data distribution directives. A pilot EPPP compiler has been implemented and preliminary results on a target distributed memory architecture (the IBM SP-1 system) are reported which demonstrate encouraging speedups of the code generated by the automatic decomposition scheme.
منابع مشابه
Automatic data and computation decomposition for distributed memory machines
In this paper, we develop an automatic compiletime computation and data decomposition technique for distributed memory machines. Our method can handle complex programs containing perfect and nonperfect loop nests with or without loop-carried dependences. Applying our decomposition algorithms, a program will be divided into collections (called clusters) of loop nests, such that data redistributi...
متن کاملOverview of Eppp -an Environment for Portable Parallel Programming
EPPP is an Environment for Portable Parallel Programming (EPPP) for current and future generation parallel computers. It is portable in the sense that the user can develop and tune his/her application on single workstations and rapidly port it and run it eeciently on a variety of parallel distributed-memory machines. In order to achieve this goal, EPPP consists of mainly four components that ar...
متن کاملSemi-automatic Process Decomposition for Non-shared Memory Machines
To achieve high performance on non·shared memory machines one must carefully distribute the data and the work so as to keep the workload balanced while minimizing the access to non-local data. Process decomposition is the operation of writing an algorithm as a collection of tasks, each operating primarily on its own portion of the data, to carry out the computation in parallel. In this paper we...
متن کاملAn Overview of the Fortran D Programming System Seema
The success of large-scale parallel architectures is limited by the diiculty of developing machine-independent parallel programs. We have developed Fortran D, a version of Fortran extended with data decomposition speciications, to provide a portable data-parallel programming model. This paper presents the design of two key components of the Fortran D programming system: a prototype compiler and...
متن کاملA Fault Diagnosis Method for Automaton based on Morphological Component Analysis and Ensemble Empirical Mode Decomposition
In the fault diagnosis of automaton, the vibration signal presents non-stationary and non-periodic, which make it difficult to extract the fault features. To solve this problem, an automaton fault diagnosis method based on morphological component analysis (MCA) and ensemble empirical mode decomposition (EEMD) was proposed. Based on the advantages of the morphological component analysis method i...
متن کامل