Abstract

This article discusses advantages and disadvantages of various computational platforms in application to problems of wave propagation simulation. Modern SIMD processors, such as multi-core ARM and Intel CPUs, as well as NVidia GPUs are considered. Wave propagation simulation is the main element of algorithms for solving inverse problems of wave tomography as coefficient inverse problems for wave equation. The field of wave tomography, which is currently under development, requires powerful computing resources. Ability to solve such problems in practice largely depends on computing hardware. The main applications of wave tomography are medical ultrasound imaging, ultrasound non-destructive testing, seismic studies. Wave simulation problem also has an applied value on its own. A convolution-type finite-difference method used in this study for benchmarking is also employed in numerous problems of mathematical physics and image processing.

中文翻译：

---

CPU和GPU平台在波浪诊断问题中的比较

摘要

本文讨论了各种计算平台在应用波传播模拟问题时的优缺点。现代 SIMD 处理器，例如多核 ARM 和 Intel CPU，以及 NVidia GPU 都被考虑在内。波传播模拟是求解波层析成像反问题的算法的主要元素，作为波动方程的系数反问题。目前正在开发的波层析成像领域需要强大的计算资源。在实践中解决此类问题的能力很大程度上取决于计算硬件。波层析成像的主要应用是医学超声成像、超声无损检测、地震研究。波浪模拟问题本身也具有应用价值。