GL Buffer with CUDA

Buffer.cu

#ifndef CUDA_GL_BUFFER
#define CUDA_GL_BUFFER

#include <glad\glad.h>
#include <cuda_runtime.h>
#include <device_launch_parameters.h>
#include <cuda_gl_interop.h>
#include <thrust\device_vector.h>
#include <thrust\host_vector.h>
#include <memory>

namespace CudaGL {

	template <typename T>
	void create_buffer(
		GLenum type, GLenum usage, cudaGraphicsRegisterFlags flags,
		std::unique_ptr <GLuint> &buffer, std::unique_ptr <cudaGraphicsResource*> &cuda_buffer,
		const thrust::host_vector<T> &data) {

		glGenBuffers(1, buffer.get());
		glBindBuffer(type, *buffer);

		glBufferData(type, data.size(), (void*)0, usage);

		cudaGraphicsGLRegisterBuffer(cuda_buffer.get(), *buffer, flags);

		send_data_to_buffer(cuda_buffer, data);
	}

	template <typename T>
	void send_data_to_buffer(std::unique_ptr <cudaGraphicsResource*> &cuda_buffer, const thrust::host_vector<T> &data) {
		thrust::device_vector<T> dev_data = data;
		T* data_ptr = thrust::raw_pointer_cast(dev_data.data());
		T* buffer;
		size_t size;

		cudaGraphicsMapResources(1, cuda_buffer.get());
		cudaGraphicsResourceGetMappedPointer((void**)&buffer, &size, *cuda_buffer);

		send_data_kernel<<<1, size >>>(buffer, data_ptr);
		cudaDeviceSynchronize();
	};

	void delete_buffer(std::unique_ptr<GLuint> &buffer, std::unique_ptr <cudaGraphicsResource*> &cuda_buffer) {
		cudaGraphicsUnregisterResource(*cuda_buffer);

		glDeleteBuffers(1, buffer.get());
	}

	template <typename T>
	__global__ void send_data_kernel(T* buffer, T* data) {
		int idx = (blockDim.x * blockIdx.x) + threadIdx.x;

		if ((*buffer)[idx] != (*data)[idx]) (*buffer)[idx] = (*data)[idx];
	}
}
#endif // !CUDA_GL_BUFFER