Abstract This article presents a benchmarking framework, namely “ProvBench,” with a specific focus on provenance of collected data, capable of identifying and measuring the impact of changes to hardware, operating system, software, middleware, and services that constitute a highly complex and heterogeneous research computing environment. The provenance is retained via detailed and automated recording of hardware details, runtime environment, software and libraries used, input data and results, as well as execution logs of the computation. This capability is particularly essential for constant monitoring and fast identification of abnormalities. The framework is compatible across different operating systems and varied software environments that support software modules. Its modular object‐oriented design allows for easy expansions, that is, adding new software and tests is straightforward. ProvBench is being actively used in our center for testing acquired equipment, evaluation of preproduction systems, assessing the impact of system and software changes, finding bad nodes, and other useful purposes with successful results.

A key trend facing extreme-scale computational science is the widening gap between computational and I/O rates, and the challenge that follows is how to best gain insight from simulation data when it is increasingly impractical to save it to persistent storage for subsequent visual exploration and analysis. One approach to this challenge is centered around the idea of in situ processing, where visualization and analysis processing is performed while data is still resident in memory. This paper examines several key design and performance issues related to the idea of in situ processing at extreme scale on modern platforms: scalability, overhead, performance measurement and analysis, comparison and contrast with a traditional post hoc approach, and interfacing with simulation codes. We illustrate these principles in practice with studies, conducted on large-scale HPC platforms, that include a miniapplication and multiple science application codes, one of which demonstrates in situ methods in use at greater than 1M-way concurrency.