Set up the CUDA Development Environment

Set up the CUDA Development Environment

To set up a CUDA development environment, in addition to installing the CUDA toolkit, you also need to install a C++ compiler. MSVC is a commonly used compiler among them.

1. Install CUDA toolkit

https://developer.nvidia.com/cuda-downloads

2. Install Visual Studio

https://visualstudio.microsoft.com/free-developer-offers

select the C++ component

If the CUDA version is old, it may not support the latest version of the Visual Studio compiler:

fatal error C1189: #error: -- unsupported Microsoft Visual Studio version! Only the versions between 2017 and 2022 (inclusive) are supported! The nvcc flag '-allow-unsupported-compiler' can be used to override this version check; however, using an unsupported host compiler may cause compilation failure or incorrect run time execution. Use at your own risk.

Try downloading a newer version of CUDA.

3. Add cl.exe to Path

If we directly use nvcc to compile C++files, an error message will appear: Cannot find compiler 'cl.exe' in PATH.

So we have to add the path of cl.exe to system variable.

G:\Program Files\Microsoft Visual Studio\2022\BuildTools\VC\Tools\MSVC\14.41.34120\bin\Hostx64\x64

Add this path to the system variable, restart your computer.

4. Using GPU to output “Hello World”

#include<stdio.h>

__global__ void hello_from_gpu()
{
    printf("Hello World from the GPU!\n");
}

int main(){
    hello_from_gpu<<<1, 1>>>();
    cudaDeviceSynchronize();
    return 0;
}

nvcc helloworld.cu

./a.exe

Reference

樊哲勇.CUDA编程基础与实践[M].北京:清华大学出版社,2020.