当安装好显卡驱动后怎么样知道驱动程序安装好了,这里以T400 + OpenEuler 正常情况下,我们只要看一下nvidia-smi 状态就可以确定他已经正常了
如图:
这里就已经确定是可以正常使用了,这里只是没有运行对应的程序,那接来下我们就写一个测试程序来测试一下:以下代码通过AI给出然后做了一些小改
这里做两个文件:
首先，让我们创建一个C文件，命名为`gpu_matrix_multiply.cu`：
#include <stdio.h>
#include <stdlib.h>
#include <cuda_runtime.h>#define N 1024 // Matrix size (N x N)
#define BLOCK_SIZE 32__global__ void matrixMultiply(float *A, float *B, float *C) {int row = blockIdx.y * blockDim.y + threadIdx.y;int col = blockIdx.x * blockDim.x + threadIdx.x;float sum = 0.0f;if (row < N && col < N) {for (int i = 0; i < N; i++) {sum += A[row * N + i] * B[i * N + col];}C[row * N + col] = sum;}
}void initMatrix(float *matrix) {for (int i = 0; i < N * N; i++) {matrix[i] = rand() / (float)RAND_MAX;}
}int main() {float *h_A, *h_B, *h_C;float *d_A, *d_B, *d_C;size_t size = N * N * sizeof(float);// Allocate host memoryh_A = (float*)malloc(size);h_B = (float*)malloc(size);h_C = (float*)malloc(size);// Initialize host matricesinitMatrix(h_A);initMatrix(h_B);// Allocate device memorycudaMalloc(&d_A, size);cudaMalloc(&d_B, size);cudaMalloc(&d_C, size);// Copy host memory to devicecudaMemcpy(d_A, h_A, size, cudaMemcpyHostToDevice);cudaMemcpy(d_B, h_B, size, cudaMemcpyHostToDevice);// Define grid and block dimensionsdim3 dimBlock(BLOCK_SIZE, BLOCK_SIZE);dim3 dimGrid((N + dimBlock.x - 1) / dimBlock.x, (N + dimBlock.y - 1) / dimBlock.y);// Create CUDA events for timingcudaEvent_t start, stop;cudaEventCreate(&start);cudaEventCreate(&stop);// Record start eventcudaEventRecord(start);// Launch kernelmatrixMultiply<<<dimGrid, dimBlock>>>(d_A, d_B, d_C);// Record stop eventcudaEventRecord(stop);cudaEventSynchronize(stop);// Calculate elapsed timefloat milliseconds = 0;cudaEventElapsedTime(&milliseconds, start, stop);printf("Matrix multiplication took %f ms\n", milliseconds);// Copy result back to hostcudaMemcpy(h_C, d_C, size, cudaMemcpyDeviceToHost);// Clean upfree(h_A); free(h_B); free(h_C);cudaFree(d_A); cudaFree(d_B); cudaFree(d_C);cudaEventDestroy(start); cudaEventDestroy(stop);return 0;
}然后能用批处理就批处理,再来创建一个Shell脚本来编译和运行这个程序。将以下内容保存为`compile_and_run.sh`：
#!/bin/bash# Compile the CUDA program
nvcc -o gpu_matrix_multiply gpu_matrix_multiply.cu# Check if compilation was successful
if [ $? -eq 0 ]; thenecho "Compilation successful. Running the program..."# Run the program./gpu_matrix_multiply
elseecho "Compilation failed."
fi然后就是跑起来:
sh compile_and_run.sh再开一个窗口来监控nvidia-smi 情况:
会看到如下结果:
这时Processes里多出来了刚才测试的程序.
测试完成.