PyCUDAにはCUDA Cが欠かせないので、CUDA Cの学習を始めることにする。取り敢えず適当なチュートリアルサイトを探して、それを叩き台にして学習することにする。

tutorial siteのgit clone¶

!mkdir git

cd git

/home/workspace/git

!git clone https://github.com/deeperlearning/professional-cuda-c-programming.git

Cloning into 'professional-cuda-c-programming'...
remote: Enumerating objects: 180, done.
remote: Counting objects: 100% (180/180), done.
remote: Compressing objects: 100% (135/135), done.
remote: Total 180 (delta 44), reused 180 (delta 44), pack-reused 0
Receiving objects: 100% (180/180), 1.22 MiB | 1.85 MiB/s, done.
Resolving deltas: 100% (44/44), done.

cd professional-cuda-c-programming

/home/workspace/git/professional-cuda-c-programming

ls

README.md  examples/  solutions/

cd examples

/home/workspace/git/professional-cuda-c-programming/examples

ls

chapter01/  chapter03/  chapter05/  chapter07/  chapter09/  common/
chapter02/  chapter04/  chapter06/  chapter08/  chapter10/

cd chapter01

/home/workspace/git/professional-cuda-c-programming/examples/chapter01

ls

Makefile  hello.cu

hello world program¶

# %load hello.cu
#include "../common/common.h"
#include <stdio.h>

/*
 * A simple introduction to programming in CUDA. This program prints "Hello
 * World from GPU! from 10 CUDA threads running on the GPU.
 */

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

int main(int argc, char **argv)
{
    printf("Hello World from CPU!\n");

    helloFromGPU<<<1, 10>>>();
    CHECK(cudaDeviceReset());
    return 0;
}

!make

nvcc -O2 -arch=sm_20 -o hello hello.cu
nvcc fatal   : Value 'sm_20' is not defined for option 'gpu-architecture'
Makefile:6: recipe for target 'hello' failed
make: *** [hello] Error 1

!nvcc -O2 -arch=sm_61 -o hello hello.cu

!./hello

Hello World from CPU!
Hello World from GPU!
Hello World from GPU!
Hello World from GPU!
Hello World from GPU!
Hello World from GPU!
Hello World from GPU!
Hello World from GPU!
Hello World from GPU!
Hello World from GPU!
Hello World from GPU!

hello world programmingは基礎中の基礎なので一応やってみた。