i Identifying Failure Modes in Compiler Algorithms applied to Distributed Memory-., Data Parallel Computation
Uložené v:
| Názov: | i Identifying Failure Modes in Compiler Algorithms applied to Distributed Memory-., Data Parallel Computation |
|---|---|
| Autori: | Peter A. Walker |
| Prispievatelia: | The Pennsylvania State University CiteSeerX Archives |
| Zdroj: | http://cs.brown.edu/research/pubs/theses/masters/1995/walker.pdf. |
| Zbierka: | CiteSeerX |
| Popis: | applied to Distributed Memory Data Parallel Computation Peter Walker This document describes the results of research examining a representative set of published algorithms used in automatic compilation of code for data parallel computation on distributed systems; we determine the data dependence conditions, as expressed in the structure of the input program, that result in these algorithms producing output code which generate incorrect result upon execution. Methods for instrumenting these compilation environments to enable run-time detection of errors in executed code are also suggested. This work examines four representative algorithms that have been developed and used in compilers for the automatic generation of code that is executed in parallel on a distributed memory architecture; in particular, all data transfer between processors is |
| Druh dokumentu: | text |
| Popis súboru: | application/pdf |
| Jazyk: | English |
| Relation: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.407.8549; http://cs.brown.edu/research/pubs/theses/masters/1995/walker.pdf |
| Dostupnosť: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.407.8549 http://cs.brown.edu/research/pubs/theses/masters/1995/walker.pdf |
| Rights: | Metadata may be used without restrictions as long as the oai identifier remains attached to it. |
| Prístupové číslo: | edsbas.E8C44100 |
| Databáza: | BASE |
Nájsť tento článok vo Web of Science | Abstrakt: | applied to Distributed Memory Data Parallel Computation Peter Walker This document describes the results of research examining a representative set of published algorithms used in automatic compilation of code for data parallel computation on distributed systems; we determine the data dependence conditions, as expressed in the structure of the input program, that result in these algorithms producing output code which generate incorrect result upon execution. Methods for instrumenting these compilation environments to enable run-time detection of errors in executed code are also suggested. This work examines four representative algorithms that have been developed and used in compilers for the automatic generation of code that is executed in parallel on a distributed memory architecture; in particular, all data transfer between processors is |
|---|