This workshop series offers a comprehensive introduction to a selection of open-source tools that enable researchers to assess the performance behaviour of their code. This course covers VTune, Maqao, and Scalasca.
With every new generation of computers, the gap between a machine’s theoretical performance and that actually delivered for real applications widens as codes struggle to exploit the hardware efficiently. It has become critical for researchers and research software engineers in HPC to understand how codes run on the hardware supplied; insight into performance can drive evolution of code and ultimately lead to advances in computational science. This workshop video series offers a comprehensive introduction to a selection of open-source tools enabling researchers to assess their code’s performance.
Originally delivered in person by Durham University’s Department of Computer Science in Q2 2023, the course is available here in a self-paced format as a collection of video presentations and workshop material.
In a landscape where computing systems are growing larger and more intricate it is critical for researchers to be able to locate areas within their code that may be causing issues or bottlenecks, yet this task is far from straightforward. This series of talks provides a structured and comprehensive guide through this complex process, clarifying the significance of performance metrics and introducing a range of tools that simplify this multifaceted process.