May 8, 2017

A Fifth Element

Krishnamoorthy adds yet another accepted paper to his PLDI scorecard

For five years now, Sriram Krishnamoorthy, System Software and Applications Capability Lead in the Advanced Computing, Mathematics, and Data Division’s High Performance Computing group, has been sharing his research involving programming languages at the Annual Association for Computing Machinery Special Interest Group on Programming Languages Conference on Programming Language Design and Implementation, known as PLDI. For PLDI 2017, Krishnamoorthy and co-author Jonathan Lifflander, from Sandia National Laboratories, will present their approach to “Cache Locality Optimization for Recursive Programs.”

In their work, Krishnamoorthy and Lifflander seek to improve application performance via cache locality optimization, where data remain in cache between consecutive uses of a data region across function boundaries to reduce reuse distance and improve memory hierarchy data reuse. They created an algorithm that generates a fork/join parallel program automatically. Then, they combined a depth-first execution of a nested fork/join program and a dependence-driven task-graph scheduler. The entire cache optimization strategy is designed to work in the context of a random work-stealing scheduler. A synopsis of their paper is now available online.

“I am so very impressed with achieving five in a row at PLDI. Each year, top-notch conferences get more and more competitive, yet Sriram has remained an almost predictable presence, showcasing his leadership and research ingenuity along the way,” said Darren Kerbyson, associate division director of PNNL’s HPC group. “His contributions continue to reflect well on the HPC work being done here at PNNL.”

PLDI is the premier conference for programming language research. This year, only 47 papers were accepted from 322 submissions, roughly a 14.5 percent success rate.

Notably, Krishnamoorthy and Lifflander co-authored an earlier paper, “Steal Tree: Low-overhead Tracing of Work Stealing Schedulers,” showcased as part of PLDI 2013, that serves as a precursor to their current cache locality optimization work. In addition, they were co-authors of “Scalable Replay with Partial-Order Dependencies for Message-Logging Fault Tolerance,” which earned the Best Student Paper Award at IEEE Cluster 2014.


  • Lifflander J and S Krishnamoorthy. 2017. “Cache Locality Optimization for Recursive Programs.” To be presented at the 38th ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI 2017), June 18-23, 2017, Barcelona, Spain.
  • Lifflander J, S Krishnamoorthy, and LV Kale. 2013. “Steal Tree: low-overhead tracing of work stealing schedulers.” In Proceedings of the 34th ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI ’13), June 16-19, 2013, Seattle, Washington. ACM, New York. DOI: 10.1145/2499370.2462193.

Download Publications

Key Capabilities


About PNNL

Pacific Northwest National Laboratory draws on its distinguishing strengths in chemistry, Earth sciences, biology and data science to advance scientific knowledge and address challenges in sustainable energy and national security. Founded in 1965, PNNL is operated by Battelle for the Department of Energy’s Office of Science, which is the single largest supporter of basic research in the physical sciences in the United States. DOE’s Office of Science is working to address some of the most pressing challenges of our time. For more information, visit For more information on PNNL, visit PNNL's News Center. Follow us on Twitter, Facebook, LinkedIn and Instagram.