A computer science professor at Colorado School of Mines is contributing to a collaborative research project that aims to pinpoint and resolve scalability issues in the software stack of multi-core and many-core hardware.
Assistant Professor Bo Wu will receive $499,891 in funding from the National Science Foundation’s Scalable Parallelism in the Extreme (SPX) program for his part of the collaborative effort with the University of Texas San Antonio (UTSA).
Unlike previous research that focused on application code to identify hidden scalability issues, the Mines and UTSA team will consider the whole software stack, including memory allocator, third-party runtime libraries and the operating system. The project’s goal is to develop analyzers to automatically pinpoint the root causes of bottlenecks and then fix them via a runtime optimizer—all without intervention from a programmer.
“The proposed techniques will greatly reduce manual effort spent on identifying scalability problems hidden in various layers of the software stack, which is expected to achieve better scalability than existing techniques in many domains, such as high-performance computing, big data analytics and machine learning,” Wu said. “The improved program performance will eventually lead to accelerated scientific discoveries and energy saving.”
At Mines, Wu’s specific focus will be developing instrumentation tools, designing runtime adaptation methods and experimenting with heterogeneous systems that consist of both central processing units (CPU) and graphics processing units (GPU).
The NSF SPX program aims to support research addressing the challenges of increasing performance in the modern era of parallel computing, which will require a collaborative effort among researchers in multiple areas, from services and applications down to micro-architecture. All projects funded through the program are collaborative in nature.