Vulkan comes to NVIDIA's CUDA thanks to the development of open source VUDA

GitHub repository called VUDA wants to bring the low-level Vulkan API to NVIDIA's CUDAs under an MIT open source license.

The video game market is one of the most important in computing, and a good number of graphics engines can be found, but in the case of APIs, only DirectX 12 was available until Vulkan arrived. Vulkan's solution is lighter and more complete, and is especially widely used in AMD graphics cards. Now, a GitHub developer named Jgbit has begun implementing something very ambitious: a project called VUDA that is based on NVIDIA's CUDA API to create a GPU computing interface accessible to everyone thanks to its open-source code.

VUDA, the Vulcan implementation in the NVIDIA CUDA API

The aim is for VUDA to develop a simple solution for the Vulkan graphics API but with low-level support for NVIDIA hardware. VUDA has a C ++ library only for the header that makes it compatible with all platforms that natively use a C ++ compiler and are also compatible with Vulkan.

Only the first steps have been taken on this project, but the community has quickly seen the potential and became interested in this open source solution under the MIT license. Inside the GitHub repository we find a very basic sample of the code that can be used as a basis for the first development with the library.

CUDA is a computing platform through parallelization and a programming model, it is not an API as is usually thought. This solution makes use of the GPU for simple and elegant cutting computing tasks. For CUDA C, C ++, Fortran and many other languages ​​are usually used, which are being incorporated and extensions are also implemented in the form of basic keywords. Keywords are what allow the developer to implement the computational parallels and even direct the compiler to the part of the application that is assigned to the GPU.