ROCm: Difference between revisions
(Created page with "ROCm is a C++ framework for GPGPU. It is AMD's attempt a CUDA.<br> [https://rocm.github.io ROCm github page]<br> [https://www.reddit.com/r/Amd/comments/a9tjge/amd_rocm_hcc_pro...") |
No edit summary |
||
| Line 8: | Line 8: | ||
* [https://www.reddit.com/r/Amd/comments/a9tjge/amd_rocm_hcc_programming_introduction/ ROCm introduction] | * [https://www.reddit.com/r/Amd/comments/a9tjge/amd_rocm_hcc_programming_introduction/ ROCm introduction] | ||
* [https://rocm.github.io/languages.html A table to transition between CUDA, OpenCL, ROCm] | * [https://rocm.github.io/languages.html A table to transition between CUDA, OpenCL, ROCm] | ||
[[Category:GPU Programming languages]] | |||
Latest revision as of 16:47, 31 January 2022
ROCm is a C++ framework for GPGPU. It is AMD's attempt a CUDA.
ROCm github page
ROCm Introduction
The main benefit over OpenCL is that your GPU code is compiled at the same time as your C++ code and is intercompatible.
You do not need to rewrite all of your C++ classes as C-style structs as you would in OpenCL.