I am trying to implement an encryption library to GPU. I am using win10, VS2017, and CUDA 9.1.
The gmp library that it uses is not allowed to call on device code.
So I have to convert some functions into device code.
Since it will take a very long time or “impossible” to convert the whole gmp library to device code, I am searching for arbitrary precision arithmetic library that I can use with CUDA.
But unfortunately, the libraries that people say like: gpuprec, cump, campary, and xmp are all linux libraries using makefile to compile which I can’t use it with visual studio.
Even though I compiled it on linux, I am unable to link the .a file and .so file on Visual Studio.
Back to the title, Is there any arbitrary precision library for CUDA using VISUAL STUDIO or some solutions that allow me to install the mentioned libraries above on VISUAL STUDIO?
I am aware that XMP have a project file and I have tried it.
I cannot compile the sample code with nvcc which show an error that it cannot read the “xmp.h” file even though I include the whole dir path like this(include “c:.…\xmp.h”).
And I also have looked up the error online. The possible reason it is not working is that I haven’t linked the library.
Then as previously mentioned, I managed to use the “makefile” to generate the library on Linux which is a “.so” and “.a” file which is not compatible with windows.
Do you have any other suggestion? or am I doing it wrong?
Thanks for your help.