Computer Science Events
This calendar presented by
Computer Science
[PAST EVENT] Colloquium: Data Locality Enhancement on GPU and Multicore
February 8, 2012
8am - 8:50am
Recent years have seen some exciting advances in computer architecture, including the rapid increase of parallelism, the rise of heterogeneity and the diversity of memory devices. These advances create tremendous computing potential, as well as new complexities for the maximization of computing efficiency. My research addresses these issues by uncovering the implications of modern hardware features, and providing compiler and run-time support to translate them into parallel software performance.
This talk will focus on the removal of irregularities in both control flows and memory references to maximize the computing efficiency of General Purpose Graphics Processing Units (GPGPU) applications. As a massively parallel many-core architecture, GPGPU is especially sensitive to such irregularities. This talk introduces a framework named G-Streamline, which provides a unified solution to both problems. It enables on-the-fly detection and elimination of irregularities through dynamic thread-data remapping, a CPU-GPU pipelining, and adaptive kernel splitting.
In addition, I will briefly talk about my other work in shared-cache-aware transformations for enhancing data locality on multicore and input-adaptive GPU code generation.
Zheng (Eddy) Zhang is a fifth-year Ph.D. student in computer science at the College of William & Mary. She received her M.S. in computer science at William and Mary with a Computational Operations Research (COR) specialization. Her research lies in the area of compilers and programming systems, with a focus on revealing and exploiting the implications of emerging hardware features on the development, compilation and execution of software. She is the lead author of a paper that wins the Best Paper Award at PPoPP '10 and is a recipient of the Google Anita Borg Memorial Scholarship.
This talk will focus on the removal of irregularities in both control flows and memory references to maximize the computing efficiency of General Purpose Graphics Processing Units (GPGPU) applications. As a massively parallel many-core architecture, GPGPU is especially sensitive to such irregularities. This talk introduces a framework named G-Streamline, which provides a unified solution to both problems. It enables on-the-fly detection and elimination of irregularities through dynamic thread-data remapping, a CPU-GPU pipelining, and adaptive kernel splitting.
In addition, I will briefly talk about my other work in shared-cache-aware transformations for enhancing data locality on multicore and input-adaptive GPU code generation.
Zheng (Eddy) Zhang is a fifth-year Ph.D. student in computer science at the College of William & Mary. She received her M.S. in computer science at William and Mary with a Computational Operations Research (COR) specialization. Her research lies in the area of compilers and programming systems, with a focus on revealing and exploiting the implications of emerging hardware features on the development, compilation and execution of software. She is the lead author of a paper that wins the Best Paper Award at PPoPP '10 and is a recipient of the Google Anita Borg Memorial Scholarship.
Contact
Department of Computer Science