I'm sorry to repeat this question, but I haven't found an answer that works yet.
I've installed CUDA 7.0.
nvcc is at C:\Apps\cuda_v7.0\toolkit\bin\nvcc.exe
CUDA_PATH = C:\Apps\cuda_v7.0\toolkit
CUDA_PATH_V7_0 = C:\Apps\cuda_v7.0\toolkit
Path includes C:\Apps\cuda_v7.0\toolkit\bin
I've downloaded CudafyV1.29 and unzipped it into C:\Projects\cudafy\CudafyV1.29
When I run C:\Projects\cudafy\CudafyV1.29\bin\CudafyViewer.exe
and 'Check for CUDA' it says
Found 1 devices.
Name: GeForce GT 555M
Compute capability: 2.1
When I press 'Run CUDA Test', it prints
Platform=x64
Checking for CUDA SDK at nvcc
CUDA SDK Version=
Attempting to cudafy a kernel function.
CUDA directory not found.
When installing CUDA, I didn't select 'CUDA Visual Studio Integration'.
I'm using Visual Studio community 2015, which has C# and C++ installed. I can't find cl.exe anywhere, though. (That's mentioned in some responses to this problem, but that's not the error I'm seeing.)
I've looked at the manual at http://www.hybriddsp.com/cudafy/CUDAfy_User_Manual_1_22.pdf and don't yet see my mistake. I've rebooted more than once. I can run nvcc --help and see lots of output. I haven't tried compiling any of nvidia's examples.
It would be helpful if CudafyViewer was more explicit in the messages that it prints, for instance, by printing the name of the environment variable it uses to look for nvcc.
Any suggestions? Thanks.
Comments: ** Comment from web user: gordonpagan **
I've installed CUDA 7.0.
nvcc is at C:\Apps\cuda_v7.0\toolkit\bin\nvcc.exe
CUDA_PATH = C:\Apps\cuda_v7.0\toolkit
CUDA_PATH_V7_0 = C:\Apps\cuda_v7.0\toolkit
Path includes C:\Apps\cuda_v7.0\toolkit\bin
I've downloaded CudafyV1.29 and unzipped it into C:\Projects\cudafy\CudafyV1.29
When I run C:\Projects\cudafy\CudafyV1.29\bin\CudafyViewer.exe
and 'Check for CUDA' it says
Found 1 devices.
Name: GeForce GT 555M
Compute capability: 2.1
When I press 'Run CUDA Test', it prints
Platform=x64
Checking for CUDA SDK at nvcc
CUDA SDK Version=
Attempting to cudafy a kernel function.
CUDA directory not found.
When installing CUDA, I didn't select 'CUDA Visual Studio Integration'.
I'm using Visual Studio community 2015, which has C# and C++ installed. I can't find cl.exe anywhere, though. (That's mentioned in some responses to this problem, but that's not the error I'm seeing.)
I've looked at the manual at http://www.hybriddsp.com/cudafy/CUDAfy_User_Manual_1_22.pdf and don't yet see my mistake. I've rebooted more than once. I can run nvcc --help and see lots of output. I haven't tried compiling any of nvidia's examples.
It would be helpful if CudafyViewer was more explicit in the messages that it prints, for instance, by printing the name of the environment variable it uses to look for nvcc.
Any suggestions? Thanks.
Comments: ** Comment from web user: gordonpagan **
I am new to this but found I had a similar problem to what is reported here. I bought a new toshiba Satellite laptop with a NVidia card with 256 CUDA cores. Install Visual Studio 2015 and then compiled up compiled up successfully CudafyExamples and CudafyByExample. Unfortunately if you try running it, it has trouble finding the CUDA directory. If you run the Cuda Viewer application as the other user said, there is a problem when your press the button 'Run CUDA Test'. It comes up with 'CUDA directory not found'.
What must I do to rectify this problem please?