CODEWITHSUNDEEP
c++cwindowscudavisual-studio-2012
Guilherme Gasparetto
Guilherme Gasparetto

Reputation: 53

Load DLL at runtime

I'm using the CUDA Driver API in a project. When I execute the project in a PC that doesn't have a NVIDIA GPU, it gives "nvcuda.dll was not found".

The problem is: this DLL is only distributed with the GPU driver, not like in the Runtime API where you can put the DLL needed with your executable. I need to load this DLL in runtime, and if it doesn't exist I will know that CUDA is not available. I'm using Visual Studio 2012 Professional.

Is there a way to do that?

Upvotes: 1

Views: 4363

Answers (2)

Robert Crovella
Robert Crovella

Reputation: 151799

Windows provides an API function (LoadLibrary) to load DLLs into memory at runtime. You provide a LPCTSTR (null terminated pointer to a const TCHAR) containing the name/path of the DLL you want to load. If you provide a relative path, Windows will scan PATH and the executable's current directory for the file. If you provide an absolute path, Windows will use that.

If LoadLibrary returns NULL, Windows couldn't find the file.

Upvotes: 1

Michael Shmalko
Michael Shmalko

Reputation: 710

Create a small "Launcher" app that will check if system meets your requirements and will launch the main application or display an error depending on the check results. To check if a DLL is available you can use LoadLibrary() as previously suggested.

Upvotes: 0

Related Questions