Set up cmake to compile library

CMakeLists.txt

@@ -1,35 +1,36 @@
-cmake_minimum_required(VERSION 3.12)
+cmake_minimum_required(VERSION 3.17)
 
-project(CUDANet)
+project(CUDANet
-
+        LANGUAGES CXX CUDA
-# Find CUDA
-find_package(CUDA REQUIRED)
-
-# Add CUDA include directories
-include_directories(${CUDA_INCLUDE_DIRS})
-
-# Add project source files
-set(SOURCES
-    src/main.cpp
-    src/utils/cuda_helper.cpp
 )
 
-# Set CUDA architecture (change according to your GPU)
+find_package(CUDAToolkit REQUIRED)
+include_directories(${CUDAToolkit_INCLUDE_DIRS})
+
+# Add project source files for the library
+set(LIBRARY_SOURCES
+    src/utils/cuda_helper.cpp
+    src/layers/dense.cpp
+)
+
+set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
 set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} -arch=sm_75)
 
-# Build executable
+# Build static library
-cuda_add_executable(${PROJECT_NAME} ${SOURCES})
+add_library(${PROJECT_NAME} STATIC ${LIBRARY_SOURCES})
 
-# Link cuBLAS library
+# Link cuBLAS library to the library
-target_link_libraries(${PROJECT_NAME} ${CUDA_cublas_LIBRARY})
+target_link_libraries(${PROJECT_NAME} CUDA::cublas CUDA::cudart)
 
-# Set include directories
+# Set include directories for the library
-target_include_directories(${PROJECT_NAME} PRIVATE
+target_include_directories(${PROJECT_NAME} PUBLIC
     ${CMAKE_CURRENT_SOURCE_DIR}/include
     ${CMAKE_CURRENT_SOURCE_DIR}/include/utils
     ${CMAKE_CURRENT_SOURCE_DIR}/include/layers
     ${CMAKE_CURRENT_SOURCE_DIR}/src
 )
 
-# Set C++ standard
+set_property(TARGET ${PROJECT_NAME} PROPERTY CXX_STANDARD 14)
-set_property(TARGET ${PROJECT_NAME} PROPERTY CXX_STANDARD 11)
+
+# Add testing subdirectory
+add_subdirectory(test)

README.md

@@ -1 +1,5 @@
 # CUDANet
+
+requirements:
+- CUDA, cuBLAS
+- Google Test 

include/layers/dense.h

@@ -15,8 +15,8 @@ namespace Layers {
         ~Dense();
 
         void forward(const float* input, float* output);
-        virtual void setWeights(const std::vector<std::vector<float>>& weights) = 0;
+        void setWeights(const std::vector<std::vector<float>>& weights);
-        virtual void setBiases(const std::vector<float>& biases) = 0;
+        void setBiases(const std::vector<float>& biases);
 
     private:
         int inputSize;

include/layers/ilayer.h

@@ -3,6 +3,7 @@
 #define I_LAYER_H
 
 #include <cublas_v2.h>
+#include <vector>
 
 namespace Layers {
 
@@ -11,6 +12,8 @@ namespace Layers {
         virtual ~ILayer() {}
 
         virtual void forward(const float* input, float* output) = 0;
+        virtual void setWeights(const std::vector<std::vector<float>>& weights) = 0;
+        virtual void setBiases(const std::vector<float>& biases) = 0;
     };
 
 } // namespace Layers

src/CMakeLists.txt

@@ -0,0 +1,6 @@
+set(LAYER_SOURCES layers/dense.cpp)
+
+add_library(CUDANet
+    utils/cuda_helper.cpp
+    ${LAYER_SOURCES}
+)

src/layers/dense.cpp

@@ -1,6 +1,7 @@
 #include "dense.h"
 #include "cuda_helper.h"
 #include <cstdlib>
+#include <cuda_runtime.h>
 #include <cublas_v2.h>
 #include <cstdio>
 #include <stdexcept>

src/main.cpp

@@ -1,68 +0,0 @@
-#include <stdio.h>
-#include <stdlib.h>
-#include <cuda_runtime.h>
-#include "cublas_v2.h"
-#include "cuda_helper.h"
-
-int main() {
-    // Initialize CUDA and get device properties
-    cudaDeviceProp deviceProp = initializeCUDA();
-
-    // Specify vector size
-    const int N = 5;
-
-    // Host vectors
-    float *h_A, *h_B, *h_C;
-
-    // Allocate host memory
-    h_A = (float*)malloc(N * sizeof(float));
-    h_B = (float*)malloc(N * sizeof(float));
-    h_C = (float*)malloc(N * sizeof(float));
-
-    // Initialize host vectors
-    for (int i = 0; i < N; ++i) {
-        h_A[i] = static_cast<float>(i);
-        h_B[i] = static_cast<float>(2 * i);
-    }
-
-    // Allocate device memory
-    float *d_A, *d_B, *d_C;
-    cudaMalloc((void**)&d_A, N * sizeof(float));
-    cudaMalloc((void**)&d_B, N * sizeof(float));
-    cudaMalloc((void**)&d_C, N * sizeof(float));
-
-    // Copy host vectors to device
-    cudaMemcpy(d_A, h_A, N * sizeof(float), cudaMemcpyHostToDevice);
-    cudaMemcpy(d_B, h_B, N * sizeof(float), cudaMemcpyHostToDevice);
-
-    // Create cuBLAS handle
-    cublasHandle_t handle;
-    cublasCreate(&handle);
-
-    // Perform vector addition: C = A + B
-    const float alpha = 1.0f;
-    const float beta = 1.0f;
-    cublasSaxpy(handle, N, &alpha, d_A, 1, d_B, 1);
-    cublasSaxpy(handle, N, &beta, d_B, 1, d_C, 1);
-
-    // Copy result from device to host
-    cudaMemcpy(h_C, d_C, N * sizeof(float), cudaMemcpyDeviceToHost);
-
-    // Display result
-    printf("Result: ");
-    for (int i = 0; i < N; ++i) {
-        printf("%f ", h_C[i]);
-    }
-    printf("\n");
-
-    // Clean up
-    free(h_A);
-    free(h_B);
-    free(h_C);
-    cudaFree(d_A);
-    cudaFree(d_B);
-    cudaFree(d_C);
-    cublasDestroy(handle);
-
-    return 0;
-}