The program of the cuda device cannot be loaded. Error (CUDA_ERROR_INVALID_IMAGE)

Error (CUDA_ERROR_INVALID_IMAGE) occurs in the following places though the program of the device was executed from the program of the host side referring to CUDA Code Samplesl (Matrix Multiplication (Driver Version)
). Why does this error occur?
Please advise.
cutilDeviceInitDrv(cuDevice, argc, argv);
CUresult status = cuCtxCreate( &cuContext, 0, cuDevice );
if ( CUDA_SUCCESS != status )
goto Error;
module_path = cutFindFilePath(“CTMerge.cubin”, argv[0]);
  if (module_path == 0) {
status = CUDA_ERROR_NOT_FOUND;
goto Error;
}
//--------------- There is no error up to here.
status = cuModuleLoad(&cuModule, module_path); //<—It makes an error here.
if ( CUDA_SUCCESS != status ) {
goto Error;

Additional information:

The binary data had adhered to the content of my cubin before and behind the text file

though the cubin file of the demonstration data was only a text file when the content of cubin was checked.

This site of CUDA build rule though I am using VC8 compile to link. (Simple CUDA build rule for Visual Studio 2005)

It twines, it downloads, and it uses it.

Is there a problem in how etc. to set this rule or is the problem caused by the difference of the Cuda version?

I am using VC8 compile to link Cuda program.

In that case, I downloaded this from (Simple CUDA build rule for Visual Studio 2005) of this site though Cuda Build rule

was necessary. Is there a problem only by this Build rule?

However, some commands of Cuda Build rule are corrected and as follows is done.

Only the file name has changed because it doesn’t understand the content of the set parameter well.

CUDA\bin\nvcc.exe" -arch sm_10 -ccbin “C:/Program Files/Microsoft Visual Studio 8/VC/bin”

-Xcompiler "/EHsc /W3 /nologo /O2 /Zi /MT " -maxrregcount=32 --compile

-o Debug/data/CTMerge.cubin CTMerge.cu

By the way, I am using Cuda v2.1. Please advise.

It was a rudimentary mistake.
The method of outputting the file of custumbuild was set by mistake.
It set to (CUDA Build Rule->General->NVCC Compilation Type ã‚’ Generate .cubin file(-cubin)) and it lived delicious.
It fired.