Prepare
If you are using CUDA for the first time, you can use the following command under Linux to check whether the CUDA compiler is installed correctly:
$ which nvcc
Typically, the command output is:
/usr/local/cuda/bin/nvcc
In addition, you may also need to check the GPU model on your machine, you can use the following command to query:
$ ls -l /dev/nv*
Possible outputs are:
crw-rw-rw- 1 root root 195, 0 Jul 3 13:44 /dev/nvidia0
crw-rw-rw- 1 root root 195, 1 Jul 3 13:44 /dev/nvidia1
crw-rw-rw- 1 root root 195, 255 Jul 3 13:44 /dev/nvidiactl
crw-rw---- 1 root root 10, 144 Jul 3 13:39 /dev/nvram
The above output shows that there are two GPU graphics cards installed on the machine.
The basic process of writing a CUDA program is:
- Create a source file, suffixed with ".cu".
- Compile the program with nvcc.
- run from the command line.
Then there is what we call the kernel function (i.e. CUDA code)
__global__ void helloFromGpu(void){
printf(“hello world form GPU!\\n”);
}
The qualifier __global__ tells the compiler that this function will be called by the CPU to execute on the GPU, and its call is of the form:
helloFromGPU<<