CUDA Compile issue : nvcc fatal : A single input...


I am trying to compile a CUDA implementation of SIFT and I keep getting the error :

nvcc fatal : A single input file is required for a non-link phase when an outputfile is specified

After going through the forums I added " " to the -I option parameters. This is the custom build rule.

(CUDA_BIN_PATH)\nvcc.exe -ccbin "(VCInstallDir)bin" -c -DWIN32 -D_CONSOLE -D_MBCS -DCUDA_SIFTGPU_ENABLED -Xcompiler /EHsc,/W3,/nologo,/Wp64,/O2,/MT -I"(CUDA_INC_PATH)" -I"./" -"I../../Include" -o (ConfigurationName)$(InputName).obj $(InputPath)

Still the same error persists. The CUDA implementation I am trying to compile is the one at

Has anyone used this CUDA SIFT implementation or would anyone have any idea on how to resolve this ?

Thanks in advance.


(CUDA_BIN_PATH)\nvcc.exe" -ccbin "(VCInstallDir)bin” -c -DWIN32 -D_CONSOLE -D_MBCS -DCUDA_SIFTGPU_ENABLED -Xcompiler “/EHsc,/W3,/nologo,/Wp64,/O2,/MT” -I"(CUDA_INC_PATH)" -I"./" -I"(NVSDKCUDA_ROOT)/common/Include" -o (ConfigurationName)\$(InputName).obj (InputPath)

Thanks for the build rule. I’ll try this out.


Tried the new rule. No change in the error. Has anyone tried to compile this implementation of SIFT on CUDA ?