- Home›
- Technology and Research›
- Intel Technology Journal›
- Multi-Core Software
Multi-Core Software
Methodology, Tools, and Techniques to Parallelize Large-Scale Applications: A Case Study
CONCLUSION
We conclude that advancements in threading analysis tools have made parallelization of complex applications an easier task than what it was a decade or two ago. The overhead of the required instrumentation to perform the dynamic dependence checking has become affordable on modern microprocessors. Effective summarization and filtering of data dependence violations play a key role in managing the large problem size. We also found that semi-automatic mechanisms provide crucial help in accomplishing the repetitive and error-prone task of source code changes. Moreover, we found that good software engineering practices make threading easier.
