ubuntu 17.04 cuda
博客专区 > MtrS 的博客 > 博客详情
ubuntu 17.04 cuda
MtrS 发表于4个月前
ubuntu 17.04 cuda
  • 发表于 4个月前
  • 阅读 22
  • 收藏 0
  • 点赞 0
  • 评论 0

腾讯云 技术升级10大核心产品年终让利>>>   

Install NVIDIA CUDA on Ubuntu 17.04

The official download page only have package for 16.04 and 14.04, but actually Ubuntu 17.04 can install CUDA via apt directly. https://launchpad.net/ubuntu/zesty/+source/nvidia-cuda-toolkit Install

Assume you already have NVIDIA graphic driver installed and just need CUDA. Only the following command is needed.

sudo apt-get install nvidia-cuda-dev nvidia-cuda-toolkit nvidia-nsight

NOTE: Ubuntu 17.04 use GCC6, which is not supported by nvcc , the package will install clang-3.8 (the default clang version for 17.04 is clang 4.0, they can co-exist). Compile

Compile cuda code using nvcc -ccbin clang-3.8 hello-world.cu , remember to use cu as suffix instead of c other wise you will have error like the following

nvcc warning : The ‘compute_20’, ‘sm_20’, and ‘sm_21’ architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
square.c:6:1: error: unknown type name ‘__global__’
__global__ void cube(float * d_out, float * d_in){

You can use the following code to test if you have correct installation


/*
* Example from Udacity Intro to Parallel Programming https://www.udacity.com/course/intro-to-parallel-programming--cs344
* nvcc -ccbin clang-3.8 cube.cu
*/
#include <stdio.h>

__global__ void cube(float * d_out, float * d_in){
	int idx = threadIdx.x;
	float f = d_in[idx];
	d_out[idx] = f * f * f;
}

int main(int argc, char ** argv) {
	const int ARRAY_SIZE = 64;
	const int ARRAY_BYTES = ARRAY_SIZE * sizeof(float);

	// generate the input array on the host
	float h_in[ARRAY_SIZE];
	for (int i = 0; i < ARRAY_SIZE; i++) {
		h_in[i] = float(i);
	}
	float h_out[ARRAY_SIZE];

	// declare GPU memory pointers
	float * d_in;
	float * d_out;

	// allocate GPU memory
	cudaMalloc((void**) &d_in, ARRAY_BYTES);
	cudaMalloc((void**) &d_out, ARRAY_BYTES);

	// transfer the array to the GPU
	cudaMemcpy(d_in, h_in, ARRAY_BYTES, cudaMemcpyHostToDevice);

	// launch the kernel
	cube<<<1, ARRAY_SIZE>>>(d_out, d_in);

	// copy back the result array to the CPU
	cudaMemcpy(h_out, d_out, ARRAY_BYTES, cudaMemcpyDeviceToHost);

	// print out the resulting array
	for (int i =0; i < ARRAY_SIZE; i++) {
		printf("%f", h_out[i]);
		printf(((i % 4) != 3) ? "\t" : "\n");
	}

	cudaFree(d_in);
	cudaFree(d_out);

	return 0;
}

Reference

https://www.udacity.com/course/intro-to-parallel-programming--cs344

https://medium.com/@at15/install-nvidia-cuda-on-ubuntu-17-04-823300ab7bcc

共有 人打赏支持
粉丝 30
博文 468
码字总数 351660
×
MtrS
如果觉得我的文章对您有用,请随意打赏。您的支持将鼓励我继续创作!
* 金额(元)
¥1 ¥5 ¥10 ¥20 其他金额
打赏人
留言
* 支付类型
微信扫码支付
打赏金额:
已支付成功
打赏金额: