Multiscale computing for science and engineering in the era of exascale performance.

Area of Science:

• Computational Science
• High-Performance Computing
• Multiscale Modeling

Background:

• Emerging exascale computing environments present new challenges for complex simulations.
• Developing and executing multiscale models requires specialized infrastructure and approaches.

Purpose of the Study:

• To discuss generic multiscale computing on exascale systems.
• To analyze the scaling of multiscale applications towards exascale performance.

Main Methods:

• Review of multiscale model development and simulation phases.
• Analysis of scaling strategies (weak, strong) in the context of exascale computing.

Main Results:

• Generic frameworks and software tools can facilitate multiscale computing.
• Traditional scaling approaches may not apply to all applications at exascale.
• Multi-scaling is proposed as a necessary paradigm for exascale performance.

Conclusions:

• Further development of generic frameworks is crucial for efficient multiscale computing.
• Exascale computing necessitates a shift towards multi-scaling for many applications.
• This work contributes to the theme of multiscale modeling and computing towards the exascale.