CUDA Support on AMD GPUs Now Possible On Windows

In the software development world, CUDA is the first choice of technology professionals since many years as a powerful parallel calculation platform offered by NVIDIA. However, this technology works traditionally only on NVIDIA GPUs, leaving AMD users outside. A new open source project brings an interesting solution to overcome this limitation.

The CUDA for AMD Windows project shared on GitHub includes:

• Providing CUDA compatibility on AMD Radeon graphics cards
• Ability to run CUDA-based applications in Windows operating system
• Active on AMD hardware with minimal change of existing CUDA code
• GPU acceleration scientific calculation and machine learning provide alternative way for workloads
• Open source code is open to the participation of the developer community

This project represents an important development especially for software developers and researchers using AMD GPU computers. If the CUDA-based libraries and applications work in a wider range of hardware, the GPU acceleration calculation will be more accessible.

While the project is still in early stages, it is still being developed with the support of the open source community. Software compatibility between AMD and NVIDIA can be considered a positive step in terms of solving problems, integrity of AI and scientific calculation ecosystem.