FFmpeg image convertion using CUDA

Hi

I want to confirm that how many faster using CUDA. So, I modified FFmpeg image convertion. FFmpeg image convertion is to convert RGB to YUV. I apply YUV2RGB api offerd CUDA.
So, i changed file name cpp to cu. And then i supply Custom Build Rule in this site.
But, i couldn’t build ffmpeg project.

============================================================

Output message is below
1>------ Rebuild All started: Project: dllFFMpeg, Configuration: Debug Win32 ------
1>Deleting intermediate and output files for project ‘dllFFMpeg’, configuration ‘Debug|Win32’
1>Compiling…
1>nvcc fatal : A single input file is required for a non-link phase when an outputfile is specified
1>Compiling…
1>FFMpegMediaCompEntryFn.cpp
1>Generating Code…
1>Compiling manifest to resources…
1>Linking…
1>LINK : fatal error LNK1104: cannot open file ‘.\emudebug\FFDecoder.obj’

====================

I don’t know why i didn’t build
Plz, if you know solution, come to knowledge.
I wait you answer.
Thank you.^^

“A single input file is required for a non-link phase when an outputfile is specified” this line means there is a space in the include path or some other paths. you should use '“” to content the path.

-I"…"