![]() |
ИСТИНА |
Войти в систему Регистрация |
ИСТИНА ИНХС РАН |
||
Advantages and disadvantages of various computing platforms in application to inverse problems of wave tomography are discussed. Modern SIMD-capable architectures are considered, including multi-core Intel CPUs such as Intel Cascade Lake 8240R and Intel Haswell-EP; Huawei ARM-based Kunpeng-920, AMD and NVIDIA GPUs The field of wave tomography, which is currently under development, requires powerful computing resources. Main applications of wave tomography are medical imaging, nondestructive testing, and seismic studies. Practical applications largely depend on computing hardware. Inverse problems of wave tomography are coefficient inverse problems for the wave equation. Such problems can be solved using iterative propagation-backpropagation technique, which relies on numerical simulation of wave propagation process. The results of benchmarking the wave simulation algorithm on different computing platforms are presented. Wave simulation problem also has an applied value on its own. Finite-difference time-domain (FDTD) method used in this study is also employed in various problems of mathematical physics and image processing. The developed algorithms may be used for solving direct and inverse problems of wave tomography. This work was supported by Huawei Technologies Co., Ltd. (Project No. OAA20100800391587A)