This title appears in the Scientific Report :
2015
Please use the identifier:
http://hdl.handle.net/2128/9544 in citations.
JUBE - A Flexible, Application- and Platform-Independent Environment for Benchmarking
JUBE - A Flexible, Application- and Platform-Independent Environment for Benchmarking
Automating benchmarks is important for reproducibility and hence comparability, which is the major intent when performing benchmarks. Furthermore managing different combinations of parameters is error-prone and often results in a significant amount of work especially if the parameter space gets larg...
Saved in:
Personal Name(s): | Lührs, Sebastian (Corresponding author) |
---|---|
Contributing Institute: |
Jülich Supercomputing Center; JSC |
Imprint: |
2015
|
Conference: | 4th JLESC Workshop, Bonn (Germany), 2015-12-02 - 2015-12-04 |
Document Type: |
Talk (non-conference) |
Research Program: |
Computational Science and Mathematical Methods |
Link: |
OpenAccess OpenAccess |
Publikationsportal JuSER |
Automating benchmarks is important for reproducibility and hence comparability, which is the major intent when performing benchmarks. Furthermore managing different combinations of parameters is error-prone and often results in a significant amount of work especially if the parameter space gets large. In order to alleviate these problems, we developed JUBE, which supports performing and analysing benchmarks in a systematic way. JUBE is an open source software tool, written in Python, providing command line accessible options to execute generic configuration files which describe the structure and behaviour of a specific benchmark. It allows adapting custom workflows to new architectures easily. Because of the generic layout JUBE can also be used for testing or production scenarios whenever a structured workflow generation and evaluation is needed. |