Summary: Computational Fluid Dynamics (CFD) commonly requires high computational power to solve complex fluid flow simulations. This study investigates the application of GPU acceleration in high-performance computing (HPC) systems using ANSYS Fluent, focusing on compute speed, power consumption, and cost.

Technical Insights:

• Assessment of various CPU and GPU architectures for CFD simulations.
• Found that GPU compute speeds are generally faster, although not always more cost-effective or power-efficient, particularly with different GPU models.
• Detailed examination of power performance differences between the Nvidia A100 and V100 GPUs.
• Evaluates the practicality of using multiple GPUs based on cost and power consumption metrics.

Significance: The findings provide valuable insights into the scalability and economic implications of integrating GPUs into computational workflows, particularly for complex simulations like CFD. This could influence future decisions in the design and deployment of HPC systems.