Quantifying Architectural Requirements of Contemporary Extreme-Scale Scientific Applications [chapter]

Jeffrey S. Vetter, Seyong Lee, Dong Li, Gabriel Marin, Collin McCurdy, Jeremy Meredith, Philip C. Roth, Kyle Spafford
2014 Lecture Notes in Computer Science  
As detailed in recent reports, HPC architectures will continue to change over the next decade in an effort to improve energy efficiency, reliability, and performance. At this time of significant disruption, it is critically important to understand specific application requirements, so that these architectural changes can include features that satisfy the requirements of contemporary extreme-scale scientific applications. To address this need, we have developed a methodology supported by a
more » ... t that allows us to investigate detailed computation, memory, and communication behaviors of applications at varying levels of resolution. Using this methodology, we performed a broad-based, detailed characterization of 12 contemporary scalable scientific applications and benchmarks. Our analysis reveals numerous behaviors that sometimes contradict conventional wisdom about scientific applications. For example, the results reveal that only one of our applications executes more floating point instructions than other types of instructions. In another example, we found that communication topologies are very regular, even for applications that, at first glance, should be highly irregular. These observations emphasize the necessity of measurementdriven analysis of real applications, and help prioritize features that should be included in future architectures. Category Metrics Computation Instruction mix Instruction categories and counts SIMD mix, width SIMD counts and vector widths Memory bandwidth Achieved R/W memory bandwidth / socket Reuse Distance Temporal data locality Communication Point-to-Point Frequency, volume, type, topology Collective Frequency, volume, type, operator B. Related work A considerable amount of previous work [4], [32], [24], [5], [22], [31] has characterized scientific applications using a variety of metrics and methodologies. This previous work PMBS13 Preliminary Version
doi:10.1007/978-3-319-10214-6_1 fatcat:fpqkbe5trrekbjvchtglthmxkq