Performance Analysis and Optimizations Techniques for Legacy Code Numerical Simulations
Numerical simulations used today by scientists in various disciplines, are frequently based on implementations created when the predominant computing hardware was sequential by design. In this simulations, new features are added or updated, when new discoveries are made, but the computational imple...
Guardado en:
| Autores principales: | , |
|---|---|
| Formato: | Objeto de conferencia |
| Lenguaje: | Inglés |
| Publicado: |
2020
|
| Materias: | |
| Acceso en línea: | http://sedici.unlp.edu.ar/handle/10915/104835 |
| Aporte de: |
| Sumario: | Numerical simulations used today by scientists in various disciplines, are frequently based on implementations created when the predominant computing hardware was sequential by design. In this simulations, new features are added or updated, when new discoveries are made, but the computational implementation remains unchanged, not taking advantage of modern hardware architectures. This “legacy code” study cases, presents the opportunity to create a set of techniques and tools, oriented to perform optimizations from a computational and software engineering points of view. As an example, in conjunction with an astrophysics research group, a real-world case numerical integrator optimization is presented, were these techniques were applied, showing the results obtained. |
|---|