CUDA implementation of integration rules within an hp-finite element code

García Prado, Adián
Pardo Zubiaur, David (dir.) ; Celorrio de Pablo, Ricardo (dir.)

Celorrio de Pablo, Ricardo (ponente)

Universidad de Zaragoza, Facultad de Ciencias, 2012
Departamento de Matemáticas,

Máster en Modelización Matemática, Estadística y Computación

Resumen: With the introduction in 2006 of CUDA architecture for Nvidia GPUs a new programming model borned. Large number of articles indicates that this new programming model in a new architecture achieves better performance than previous implementations in traditional languages for CPUs. In this work the author tries to show the capabilities of GPU computing. To perform such a task a hp Finite Element integration method is implemented both in CUDA and in C language. After implementation, parallel executions in CPU and GPU will be compared to demonstrate if it is worth to create new algorimths under this architecture.

Palabra(s) clave (del autor): integration ; fem ; cuda ; C language
Tipo de Trabajo Académico: Trabajo Fin de Master

Creative Commons License
Registrado por la Universidad de Zaragoza bajo la licencia Creative Commons.

Valore este documento:

Rate this document:
(Sin ninguna reseña)