| Project/Area Number |
14580386
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
計算機科学
|
|---|
| Research Institution | The university of Aizu |
|---|
Principal Investigator |
MINYI Guo Univ of Aizu, Dept.of Computer Software, Associate Professor, コンピュータ理工学部, 講師 (20332934)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
YI Pan Georgia State Univ., Dept.of Computer Science, Associate Professor, 助教授
NAKATA Ikuo Hosei Univ., Faculty of Computer and Information Sciences, Professor, 情報科学部, 教授 (70133022)
|
|---|
| Project Period (FY) |
2002 – 2003
|
| Project Status |
Completed (Fiscal Year 2003)
|
| Budget Amount *help |
¥4,000,000 (Direct Cost: ¥4,000,000)
Fiscal Year 2003: ¥1,500,000 (Direct Cost: ¥1,500,000)
Fiscal Year 2002: ¥2,500,000 (Direct Cost: ¥2,500,000)
|
|---|
| Keywords | Parallelizing Compilers / Automatic Parallelization / Irregular Scientific Computing / Data Redistribution / Loop Partitioning / Communication Optimization / Distributed memory multicomputers / International researcher interchange / ループ分割 / 通信最適化 |
|---|
| Research Abstract |
This research project mainly pursued the following new techniques and approaches for irregular scientific computing :
1.Communication optimization techniques for the irregular array references in nested loops. In our methods, the communication set is generated at compile-time by introducing symbolic analysis. Symbolic solutions of a set of symbolic expression are obtained by using certain restrictions. We introduced symbolic analysis algorithms to obtain the solutions in terms of a set of equalities and inequalities.
2.Loop partitioning for irregular parallel codes. We propose a communication cost reduction computes rule for irregular loop partitioning, called least communication computes rule. The loop iterations are partitioned to processors on which the minimal communication cost is ensured when executing that loop.
3.OpenMP directive extensions. We proposed some extended implementation of OpenMP directives, aiming at irregular computing OpenMP codes executed in parallel efficiently. These Open MP directives include scheduling for irregular loops, parallelizing irregular reduction, and eliminating ordered loops.
4.Efficient implementation for GEN_BLOCK redistribution in HPF. There is a compiler directive GEN_BLOCK in HPF-2 for irregular data redistribution to achieve load balance. Until now, the implementation of this directive is costly. This algorithm attempts to obtain near optimal scheduling while satisfying the conditions of minimal message size to total steps and the minimal number of steps for irregular array redistribution.
|
