init

.gitignore

@@ -0,0 +1 @@
+build

.vscode/extensions.json

@@ -0,0 +1,13 @@
+{
+    "recommendations": [
+        "ms-vscode.cpptools",
+        "ms-vscode.cmake-tools",
+        "josetr.cmake-language-support-vscode",
+        "nvidia.nsight-vscode-edition",
+        "IBM.output-colorizer",
+        "christian-kohler.path-intellisense",
+    ],
+    "unwantedRecommendations": [
+        "llvm-vs-code-extensions.vscode-clangd"
+    ]
+}

.vscode/settings.json

@@ -0,0 +1,21 @@
+{
+    "cmake.debugConfig": {
+        "args": [
+            // 程序运行参数
+        ]
+    },
+    "cmake.configureSettings": {},
+    "cmake.outputLogEncoding": "utf8bom",
+    "cmake.generator": "Ninja",
+    // "cmake.enableLanguageServices": false,
+    "[cpp]": {
+        "files.encoding": "utf8bom"
+    },
+    "[cuda-cpp]": {
+        "files.encoding": "utf8bom",
+    },
+    "files.associations": {
+        "iostream": "cpp",
+        "chrono": "cpp"
+    },
+}

CMakeLists.txt

@@ -0,0 +1,31 @@
+cmake_minimum_required(VERSION 3.20)
+project("WIN_CUDA_DEV")
+
+if(NOT WIN32)
+    message(FATAL_ERROR "This project only supports Windows platform")
+endif()
+
+# 语法标准
+enable_language(CXX CUDA)
+set(CMAKE_CXX_STANDARD 17)
+set(CMAKE_CUDA_STANDARD 17)
+
+# 显卡架构优化，详见 https://developer.nvidia.com/cuda-gpus
+set(CMAKE_CUDA_ARCHITECTURES 86)
+
+# 查找CUDA库
+find_package(CUDAToolkit REQUIRED)
+
+# 头文件索引
+include_directories("./include")
+
+# 添加可执行文件 test_add.exe
+add_executable(test_add
+    "./src/test_add.cpp"
+    "./src/external/main_ext.cpp"
+)
+
+# cuda-test.exe
+add_executable(test_cuda src/test_cuda.cu)
+target_link_libraries(test_cuda CUDA::cudart)
+set_target_properties(test_cuda PROPERTIES CUDA_SEPARABLE_COMPILATION ON)

README.md

@@ -0,0 +1 @@
+# Windows平台 VSCode CMake CUDA 开发

include/main_ext.h

@@ -0,0 +1,3 @@
+#pragma once
+
+int add(int,int);

src/external/main_ext.cpp

@@ -0,0 +1,6 @@
+#include <main_ext.h>
+
+auto add(int x, int y) -> int
+{
+    return x + y;
+}

src/test_add.cpp

@@ -0,0 +1,8 @@
+#include <main_ext.h>
+#include <iostream>
+int main()
+{
+    std::cout << "Windows平台 VSCode + CMake + CUDA 开发工具集" << std::endl;
+    std::cout << " 1 + 2 = " << add(1, 2) << std::endl;
+    return 0;
+}

src/test_cuda.cu

@@ -0,0 +1,112 @@
+#include <cuda_runtime.h>
+#include <device_launch_parameters.h>
+#include <iostream>
+#include <vector>
+#include <chrono>
+#include <cmath>
+
+// CUDA核函数：向量加法
+__global__ void vectorAdd(const float *A, const float *B, float *C, int N)
+{
+    int i = blockDim.x * blockIdx.x + threadIdx.x;
+    if (i < N)
+    {
+        C[i] = A[i] + B[i];
+    }
+}
+
+// 检查CUDA错误
+#define CUDA_CHECK(call)                                                                                                     \
+    do                                                                                                                       \
+    {                                                                                                                        \
+        cudaError_t error = call;                                                                                            \
+        if (error != cudaSuccess)                                                                                            \
+        {                                                                                                                    \
+            std::cerr << "CUDA error at " << __FILE__ << ":" << __LINE__ << " - " << cudaGetErrorString(error) << std::endl; \
+            exit(1);                                                                                                         \
+        }                                                                                                                    \
+    } while (0)
+
+int main()
+{
+    // 向量大小
+    const int N = 1024 * 1024;
+    const size_t size = N * sizeof(float);
+
+    // 在主机上分配内存
+    std::vector<float> h_A(N);
+    std::vector<float> h_B(N);
+    std::vector<float> h_C(N);
+
+    // 初始化向量
+    for (int i = 0; i < N; i++)
+    {
+        h_A[i] = static_cast<float>(i);
+        h_B[i] = static_cast<float>(i * 2);
+    }
+
+    // 在设备上分配内存
+    float *d_A = nullptr;
+    float *d_B = nullptr;
+    float *d_C = nullptr;
+
+    CUDA_CHECK(cudaMalloc(&d_A, size));
+    CUDA_CHECK(cudaMalloc(&d_B, size));
+    CUDA_CHECK(cudaMalloc(&d_C, size));
+
+    // 将数据从主机复制到设备
+    CUDA_CHECK(cudaMemcpy(d_A, h_A.data(), size, cudaMemcpyHostToDevice));
+    CUDA_CHECK(cudaMemcpy(d_B, h_B.data(), size, cudaMemcpyHostToDevice));
+
+    // 配置执行参数
+    int threadsPerBlock = 256;
+    int blocksPerGrid = (N + threadsPerBlock - 1) / threadsPerBlock;
+
+    // 记录开始时间
+    auto start = std::chrono::high_resolution_clock::now();
+
+    // 启动CUDA核函数
+    vectorAdd<<<blocksPerGrid, threadsPerBlock>>>(d_A, d_B, d_C, N);
+
+    // 检查核函数启动错误
+    CUDA_CHECK(cudaGetLastError());
+
+    // 等待GPU完成所有操作
+    CUDA_CHECK(cudaDeviceSynchronize());
+
+    // 记录结束时间
+    auto end = std::chrono::high_resolution_clock::now();
+    auto duration = std::chrono::duration_cast<std::chrono::microseconds>(end - start);
+
+    // 将结果从设备复制回主机
+    CUDA_CHECK(cudaMemcpy(h_C.data(), d_C, size, cudaMemcpyDeviceToHost));
+
+    // 验证结果
+    bool success = true;
+    for (int i = 0; i < N; i++)
+    {
+        if (std::abs(h_C[i] - (h_A[i] + h_B[i])) > 1e-5)
+        {
+            success = false;
+            break;
+        }
+    }
+
+    // 输出结果
+    std::cout << "CUDA Vector Addition Test:" << std::endl;
+    std::cout << "Vector size: " << N << std::endl;
+    std::cout << "Execution time: " << duration.count() << " microseconds" << std::endl;
+    std::cout << "Verification: " << (success ? "PASSED" : "FAILED") << std::endl;
+    std::cout << "Sample results (first 10 elements):" << std::endl;
+    for (int i = 0; i < 10; i++)
+    {
+        std::cout << h_A[i] << " + " << h_B[i] << " = " << h_C[i] << std::endl;
+    }
+
+    // 释放设备内存
+    CUDA_CHECK(cudaFree(d_A));
+    CUDA_CHECK(cudaFree(d_B));
+    CUDA_CHECK(cudaFree(d_C));
+
+    return 0;
+}