Accelerating numerical modeling of wave propagation through 2-D anisotropic materials using OpenCL

Area of Science:

• Geophysics
• Computational Science
• Materials Science

Background:

• Numerical modeling of elastic wave propagation is crucial for understanding seismic phenomena.
• Anisotropic materials exhibit complex wave behaviors requiring advanced simulation techniques.
• Parallel computing devices (PCDs) offer potential for accelerating computationally intensive simulations.

Purpose of the Study:

• To implement and validate a numerical model for 2D elastic wave propagation in anisotropic materials.
• To assess the performance and capabilities of emerging parallel computing devices (PCDs) for seismic wave modeling.
• To model laboratory experiments involving ultrasonic wave interaction with rotated anisotropic materials.

Main Methods:

• A finite difference code was developed for simulating elastic wave propagation in 2D anisotropic media.
• The code was accelerated using the PyOpenCL toolkit for utilizing Open Computing Language (OpenCL) APIs on PCDs.
• Simulations covered a range of rotation angles for transversely anisotropic and weakly orthorhombic materials.

Main Results:

• A significant speedup factor exceeding 19 was achieved using a Graphics Processing Unit (GPU) compared to a multi-core Central Processing Unit (CPU).
• Performance was evaluated across different graphic cards and operating systems, demonstrating the versatility of the PyOpenCL implementation.
• Numerical modeling successfully replicated transmission and reflection signals observed in laboratory experiments.

Conclusions:

• PyOpenCL provides an efficient method for accelerating elastic wave propagation simulations on PCDs.
• Emerging PCDs, particularly GPUs, show substantial promise for advancing geophysical modeling capabilities.
• The developed finite difference code and PyOpenCL implementation offer a valuable tool for both experimental modeling and hardware performance analysis.