NVIDIA CUDA编程实操一：环境检查与入门

我使用的华为云开发者的AI Gallery提供的环境，选择一个Notebook案例运行即可进入，进入后可以切换到限时免费的GPU环境。

第一阶段：环境检查

1. 验证CUDA驱动安装

nvidia-smi

输出：

Thu Jun  5 16:57:18 2025       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 470.57.02    Driver Version: 470.57.02    CUDA Version: 11.4     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  Tesla P100-PCIE...  On   | 00000000:00:0D.0 Off |                    0 |
| N/A   32C    P0    25W / 250W |      0MiB / 16280MiB |      0%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

2. 检查CUDA工具包

nvcc --version

输出：

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2017 NVIDIA Corporation
Built on Fri_Sep__1_21:08:03_CDT_2017
Cuda compilation tools, release 9.0, V9.0.176

3. 验证CUDA示例运行

cd /usr/local/cuda/samples/1_Utilities/deviceQuery
make
./deviceQuery

该环境无示例。

第二阶段：第一个CUDA程序

1. 创建hello.cu文件

#include <stdio.h>

__global__ void helloFromGPU() {
    printf("Hello World from GPU thread %d!\n", threadIdx.x);
}

int main() {
    printf("Hello World from CPU!\n");
    
    helloFromGPU<<<1, 5>>>();
    cudaDeviceSynchronize();
    
    return 0;
}

2. 编译并运行

nvcc hello.cu -o hello
./hello

预期输出：

Hello World from CPU!
Hello World from GPU thread 0!
Hello World from GPU thread 1!
Hello World from GPU thread 2!
Hello World from GPU thread 3!
Hello World from GPU thread 4!

第三阶段：设备信息查询

1. 创建device_info.cu

#include <stdio.h>

int main() {
    int deviceCount;
    cudaGetDeviceCount(&deviceCount);
    
    for (int i = 0; i < deviceCount; i++) {
        cudaDeviceProp prop;
        cudaGetDeviceProperties(&prop, i);
        
        printf("Device %d: %s\n", i, prop.name);
        printf("  Compute capability: %d.%d\n", prop.major, prop.minor);
        printf("  Global memory: %.2f GB\n", prop.totalGlobalMem/1024.0/1024.0/1024.0);
        printf("  Shared memory per block: %.2f KB\n", prop.sharedMemPerBlock/1024.0);
        printf("  Warp size: %d threads\n", prop.warpSize);
        printf("  Max threads per block: %d\n", prop.maxThreadsPerBlock);
        printf("  Max threads per multiprocessor: %d\n", prop.maxThreadsPerMultiProcessor);
        printf("  Multiprocessor count: %d\n", prop.multiProcessorCount);
    }
    
    return 0;
}

2. 编译运行

nvcc device_info.cu -o device_info
./device_info

对于Tesla P100应看到类似：

Device 0: Tesla P100-PCIE-16GB
  Compute capability: 6.0
  Global memory: 15.90 GB
  Shared memory per block: 48.00 KB
  Warp size: 32 threads
  Max threads per block: 1024
  Max threads per multiprocessor: 2048
  Multiprocessor count: 56

第四阶段：简单向量加法

1. 创建vector_add.cu

#include <stdio.h>
#include <cuda_runtime.h>

#define N 5

__global__ void vectorAdd(int *a, int *b, int *c) {
    int tid = blockIdx.x * blockDim.x + threadIdx.x;
    if (tid < N) {
        c[tid] = a[tid] + b[tid];
    }
}

int main() {
    int a[N] = {1, 2, 3, 4, 5};
    int b[N] = {10, 20, 30, 40, 50};
    int c[N] = {0};
    
    int *d_a, *d_b, *d_c;
    cudaMalloc(&d_a, N * sizeof(int));
    cudaMalloc(&d_b, N * sizeof(int));
    cudaMalloc(&d_c, N * sizeof(int));
    
    cudaMemcpy(d_a, a, N * sizeof(int), cudaMemcpyHostToDevice);
    cudaMemcpy(d_b, b, N * sizeof(int), cudaMemcpyHostToDevice);
    
    vectorAdd<<<1, N>>>(d_a, d_b, d_c);
    
    cudaMemcpy(c, d_c, N * sizeof(int), cudaMemcpyDeviceToHost);
    
    for (int i = 0; i < N; i++) {
        printf("%d + %d = %d\n", a[i], b[i], c[i]);
    }
    
    cudaFree(d_a);
    cudaFree(d_b);
    cudaFree(d_c);
    
    return 0;
}

2. 编译运行

nvcc vector_add.cu -o vector_add
./vector_add

输出：

1 + 10 = 11
2 + 20 = 22
3 + 30 = 33
4 + 40 = 44
5 + 50 = 55