Machine-Learning-based Autotuning

PI: Apan Qasem

Students: Saami Rahman, Biplab Kumar Saha, Tiffany Connors, Illiana Reed

Collaborators: Qing Yi (UCCS), Martin Burtscher (Texas State)

Funding: National Science Foundation Early CAREER Award

The goal of the proposed research is to formulate novel and efficient strategies for automatic performance tuning in order to harness the computational power of current and future highperformance architectures. Achieving a high fraction of peak performance on complex architectures has been a perennial challenge for application developers. The emergence of multicore processors and accelerators has greatly exacerbated this problem. With an increasing number of cores per socket, deep hierarchies of shared and distributed caches, and exascale computing on the horizon, multicore platforms pose unprecedented challenges for software development and application porting. This research aims to confront the challenge of multicore and manycore software development by significantly improving automatic performance tuning through improved feedback diagnostics. Efficiency is be achieved through enhanced knowledge of the problem domain, program features and architectural characteristics. To this end, we are developing and extending a set of tools that allow specification, collection and synthesis of tuning related information. We are also designing novel search heuristics and augmenting machine learning models to take advantage of this rich set of information to deliver scalable, portable and sustainable performance on diverse architectures.


CRL research is supported with generous funding from the National Science Foundation, Department of Energy, Semiconductor Research Consortium (SRC), IBM, NVidia, Rice University and the Research Enhancement Program at Texas State University.


Apan Qasem
Department of Computer Science
Texas State University
601 University Dr
San Marcos, TX 78666

Office: Comal 307A
Phone: (512) 245-0347
Fax: (512) 245-8750
E-mail: apan "AT" txstate · edu