Software maturity level methodology
Pursuant to EESI1 recommendations to build a European Exascale Software Centre EESI2 has created a maturity test model that will help validate exascale components and modules.
EESI2 has examined « equivalent » centres and propose a first model to support its operation and testing the level of maturity of software components.
Four main HPC software component classes are defined, namely scientific application codes, libraries and frameworks, development tools, and programming model implementations.
The core of the model are the assessment criteria for HPC software maturity : documentation, support, availability, coverage, portability, scalability, performance, and quality.
The maturity model can be experimented by identified project candidates.
The activities related to Education and strategic collaborations acknowledge the key role of education to build exascale expertise in Europe.
Recommendations foresee that the key players should develop credible and innovative educational strategies as integral part of current and future ExaScale
International collaboration, such as the Big Data Extreme Computing workshops, and dissemination actions address the same need to grow expertise and lead the way toward exascale computing.
ExaScale computing creates opportunities across all fields of science and technology. However, to leverage the full potential and to exploit ExaScale as primary driver for technological innovation and scientific progress, education and training are indentified as key bottleneck across all levels of the value added chain.
In the short term, the training of scientists and engineers must be improved on a technical level by strengthening existing mechanisms. These are primarily the compact training courses that are already offered by computing centers and other institutions.
The more fundamental innovation will critically depend on creating programs at the university level that aim at educating a new kind of computational scientist or computational engineer whose must build a bridge between the traditional disciplines. This requires mid-term and long term strategic planning.
EESI2 therefore recommends that the EU supports these necessary strategic actions by well-balanced incentives that are suitably integrated into the funding mechanisms. EU funding for ExaScale computing must acknowledge the key role of education by requiring that the key players develop credible and innovative educational strategies as integral part of current and future ExaScale investments. This should include the traditional training courses for their existing base of users, it must include suitable training for that part of the community that is already compute-affine.
However, the more fundamental paradigm shift of ExaScale requires also decisive action on the level of university education. The EU should require that any larger investment in hardware, software, and ExaScale related research is interfaced or integrated with ExaScale-oriented interdisciplinary educational programs at the participating institutions.