Collin McMillan and Zheng (Eddy) Zhang, both doctoral students in the Department of Computer Science, have been selected for the 2011 Stephen K. Park Graduate Research Award. This $1,000 award is presented annually in the fall semester to at least one currently registered Computer Science graduate student who in the prior academic year "established or refined, and documented, a research result judged to be particularly significant by the joint faculty of the Computer Science department, or a subcommittee thereof."
Collin was nominated by his advisor, Denys Poshyvanyk, for his result published in the paper Portfolio: Finding Relevant Functions and Their Usages, which appeared in the Proceedings of the 33rd IEEE/ACM International Conference on Software Engineering (ICSE'11). In his nomination letter, Denys notes that the most significant contribution of this paper is in the new ranking mechanism for source code search and retrieval in large C++ source code repositories. The proposed algorithm is based on a combination of natural language processing (NLP), the PageRank link analysis algorithm, and the spreading activation method for searching networks. Particularly novel is the use of SNA since it appears that the associated functions that are not directly hit by the keywords are ranked higher and thus become easy to retrieve. Portfolio was evaluated in a large-scale user study, which demonstrated that Portfolio was able to find relevant results with higher accuracy when compared to Google Code Search and Koders, publicly available search engines for open source code.
Eddy was nominated by her advisor, Xipeng Shen, for the result that first appeared as Does Cache Sharing on Modern CMP Matter to the Performance of Contemporary Multithreaded Programs? in the proceedings of the 15th ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming (PPoPP '10), which won a Best Paper Award, and most recently appeared in the journal paper The Significance of CMP Cache Sharing on Contemporary Multithreaded Applications in the IEEE Transactions on Parallel and Distributed Systems. In his nomination letter, Xipeng notes that Eddy conducted the first systematic measurement of the influence of cache sharing on two kinds of commodity chip multiprocessor (CMP) machines and revealed some surprising results: Contrary to the commonly perceived importance of cache sharing, neither positive nor negative effects from cache share are significant for most programs, regardless of the types of parallelism, the sets of input data, the architectures, the numbers of threads, and the assignments of threads to cores. After a detailed analysis, Eddy uncovered the main reason and invented a new paradigm for locality enhancement, cache-sharing-aware transformations, and demonstrated performance increases of up to 36% on a set of standard benchmarks.
Both Collin and Eddy have received other recognition for their work: Collin is the recipient of a NASA Virginia Space Grant Consortium 2011-2012 Graduate STEM Research Fellowship and Eddy is the first W&M student to receive a Google Anita Borg Memorial Scholarship. Both expect to defend their dissertations this spring and both are currently searching for positions.
To learn more about Steve Park, a former faculty member and chair of the department whose bequest endowed this award, visit the commemoration on our Facebook page.