Error compiling CUDA from Command Prompt

53
votes

I'm trying to compile a cuda test program on Windows 7 via Command Prompt, I'm this command:

nvcc test.cu

But all I get is this error:

nvcc fatal : Cannot find compiler 'cl.exe' in PATH

What may be causing this error?

6
cudanvidia
if you get another error "Cannot find corecrt.h" after fixing this issue, check out stackoverflow.com/questions/38290169/… - Alexander Pacha

6 Answers

54
votes

You will need to add the folder containing the "cl.exe" file to your path environment variable. For example: 

C:\Program Files\Microsoft Visual Studio 10.0\VC\bin

Edit: Ok, go to My Computer -> Properties -> Advanced System Settings -> Environment Variables. Here look for "PATH" in the list, and add the path above (or whatever is the location of your cl.exe).

27
votes

For new Visual Studio cl.exe is present in path => C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\14.23.28105\bin\Hostx64\x64

x64 is for 64bit

x86 is for 32bit

21
votes

Solve this problem by adding this options to nvcc

nvcc x.cu ...   -ccbin "D:\Program Files\Microsoft Visual Studio 11.0\VC\bin"

for example my compiler is VS2012. and cl.exe is in this dir

4
votes

cl.exe is Microsoft's C/C++ compiler. So the problem is that you don't have that installed where the command line can find it.

1
votes

nvcc is only a front end for the CUDA specific part of the program. It must invoke a full compiler to finish the job. In this case it cannot find the Visual Studio compiler 'cl.exe'

Check paths, nvcc documentation etc.

0
votes

I see that this is an old question but I recently got this error on my Visual Studio 2012 when I tried to build my CUDA project. Apparently I had changed my CUDA project to the Nov 2012 pack, changing it back to the v110 that it usually is by default fixed this error.

In Visual Studio, left click on the CUDA project, ->properties->Configuration Properties-> General -> Platform toolset, and choose: Visual Studio 2012 (v110).

I could probably get it to work with the Nov 2012 pack, but the CUDA code does not use any of the additional functions of that pack, so it is not necessary. (That pack contains the variadic templates for C++11.)