Using OpenCL technology for structured grid computing using GPU
Бесплатный доступ
In this article the possibility of using OpenCL technology for parallel calculation of a numerical solution using the example of a two-dimensional heat conduction problem is discussed. The features of programming in the paradigm of task and data parallelism are considered. The standard difficulties and methods built into OpenCL for solving them when building a program for calculations on multidimensional grids are shown. For the CPU, a similar single-threaded program was implemented in C++20, and the performance was compared.
Opencl
Короткий адрес: https://sciup.org/147250350
IDR: 147250350
Статья научная