2019 Graduate Student Showcase
 

Degree Program

Computer Science, MS

Major Advisor Name

Catherine Olschanowsky

Type of Submission

Scholarly Poster

Abstract

Achieving high application performance depends on the combination of memory footprint, instruction mix and order, and memory access patterns. Most memory benchmarks which provide information on the achieved memory performance are confined to simple access patterns that are not representative of patterns found in real applications.

We present AdaptMemBench, a configurable benchmark framework designed to explore the performance capabilities of compute kernels extracted from applications. AdaptMemBench provides a framework to emulate application-specific memory access patterns. The build system accommodates the polyhedral model, which provides a convenient testbed for code optimizations. AdaptMemBench supports reproducibility in experimental results and facilitates sharing results.

Share

COinS