|
| This Article | ||
| ||
| Share | ||
| Bibliographic References | ||
| Add to: | ||
 Digg  Furl  Spurl  Blink  Simpy  Google  Del.icio.us  Y!MyWeb | ||
| Search | ||
| ||
11th International Conference on Parallel Architectures and Compilation Techniques (PACT'02)
An Evaluation of Data-Parallel Compiler Support for Line-Sweep Applications
Charlottesville, Virginia
September 22-September 25
ISBN: 0-7695-1620-3
| ASCII Text | x | ||
| Daniel Chavarría-Miranda, John Mellor-Crummey, "An Evaluation of Data-Parallel Compiler Support for Line-Sweep Applications," Parallel Architectures and Compilation Techniques, International Conference on, pp. 7, 11th International Conference on Parallel Architectures and Compilation Techniques (PACT'02), 2002. | |||
| BibTex | x | ||
| @article{ 10.1109/PACT.2002.1105969, author = {Daniel Chavarría-Miranda and John Mellor-Crummey}, title = {An Evaluation of Data-Parallel Compiler Support for Line-Sweep Applications}, journal ={Parallel Architectures and Compilation Techniques, International Conference on}, volume = {0}, year = {2002}, issn = {1089-795X}, pages = {7}, doi = {http://doi.ieeecomputersociety.org/10.1109/PACT.2002.1105969}, publisher = {IEEE Computer Society}, address = {Los Alamitos, CA, USA}, } | |||
| RefWorks Procite/RefMan/Endnote | x | ||
| TY - CONF JO - Parallel Architectures and Compilation Techniques, International Conference on TI - An Evaluation of Data-Parallel Compiler Support for Line-Sweep Applications SN - 1089-795X SP EP A1 - Daniel Chavarría-Miranda, A1 - John Mellor-Crummey, PY - 2002 KW - null VL - 0 JA - Parallel Architectures and Compilation Techniques, International Conference on ER - | |||
Data parallel compilers have long aimed to equal the performance of carefully and-optimized parallel codes. For tightly-coupled applications based on line sweeps, this goal has been particularly elusive. In the Rice dHPF compiler, we have developed a wide spectrum of optimizations that enable us to closely approach hand-coded performance for tightly-coupled line sweep applications including the NAS SP and BT benchmark codes. From lightly-modified copies of standard serial versions of these benchmarks, dHPF generates MPI-based parallel code that is within 4% of the performance of the hand-crafted MPI implementations of these codes for a 1023 problem size (Class B) on 64 processors.We describe and quantitatively evaluate the impact of partitioning, communication and memory hierarchy optimizations implemented by dHPF that enable us to approach hand-coded performance with compiler-generated parallel code.
Citation:
Daniel Chavarría-Miranda, John Mellor-Crummey, "An Evaluation of Data-Parallel Compiler Support for Line-Sweep Applications," pact, pp.7, 11th International Conference on Parallel Architectures and Compilation Techniques (PACT'02), 2002
Usage of this product signifies your acceptance of the Terms of Use.