cudaMemcpy2DFromArray resulting in cudaErrorInvalidValue

Accelerated Computing CUDA CUDA Setup and Installation
1

I am having some troubles using OpenGL-CUDA-Interopt
My setup is the following:

I am using a FrameBufferObject for render to texture (this works very well) which I am creating this way:

void createFrameBuffer(){
    
    	// Create the texture for the FBO
    	glGenTextures(1, &tex_data);
    	glBindTexture(GL_TEXTURE_2D, tex_data);
    	SDK_CHECK_ERROR_GL();
    	glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
        glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
        glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
        glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
    
    	glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA16F , window_w, window_h, 0, GL_RGBA, GL_FLOAT, NULL);
    
    	// Create the frameBuffer
    	glGenFramebuffersEXT(1, &fbo);
        glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, fbo);
    	// Register the texture with cuda
    	checkCudaErrors(cudaGraphicsGLRegisterImage(&res_data, tex_data, GL_TEXTURE_2D, cudaGraphicsMapFlagsReadOnly));
    	SDK_CHECK_ERROR_GL();
        // attach images
    
    	glGenRenderbuffersEXT(1, &depth_rb);
    	glBindRenderbufferEXT(GL_RENDERBUFFER_EXT, depth_rb);
    	glRenderbufferStorageEXT(GL_RENDERBUFFER_EXT, GL_DEPTH_COMPONENT24, window_w, window_h);
    	//-------------------------
    
    
        glFramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, GL_COLOR_ATTACHMENT0_EXT, GL_TEXTURE_2D, tex_data, 0);
    	   //Attach depth buffer to FBO
        glFramebufferRenderbufferEXT(GL_FRAMEBUFFER_EXT, GL_DEPTH_ATTACHMENT_EXT, GL_RENDERBUFFER_EXT, depth_rb);
        //glFramebufferRenderbufferEXT(GL_FRAMEBUFFER_EXT, GL_COLOR_ATTACHMENT0_EXT, GL_RENDERBUFFER_EXT, color);
        //glFramebufferRenderbufferEXT(GL_FRAMEBUFFER_EXT, GL_DEPTH_ATTACHMENT_EXT, GL_RENDERBUFFER_EXT, depth);
    
        // clean up
        glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, 0);
    	SDK_CHECK_ERROR_GL();
    }

Currently I am not having a kernel, but I want to convert the texture from OpenGL to something that I can use with CUDA. Here is what I am trying for conversion:

void processImage(){
    	cudaArray *cuda_data = 0;
    	checkCudaErrors( cudaGraphicsGLRegisterImage(&res_data, tex_data, GL_TEXTURE_2D, cudaGraphicsRegisterFlagsReadOnly));
    	float *dData; // Array holding the data
    	std::cout << "cudaMalloc " << std::endl;
    	checkCudaErrors( cudaMalloc(&dData, window_w * window_h*4*sizeof(float)));   
    	
    	std::cout << "Mapping Resource" << std::endl;
    	checkCudaErrors( cudaGraphicsMapResources(1, &res_data) );
    
    	std::cout << "cudaGraphicsSubResourceGetMappedArray" << std::endl;
    	checkCudaErrors( cudaGraphicsSubResourceGetMappedArray(&cuda_data, res_data, 0, 0));
    
    	std::cout << "cudaMemcpy2DFromArray" <<  std::endl;
    	// Copy from the texture to the graphic Card
    	checkCudaErrors( cudaMemcpy2DFromArray(dData, window_w*4*sizeof(float), cuda_data ,0 ,0, window_w*4*sizeof(float), window_h, cudaMemcpyDeviceToDevice));
    	
    	std::cout << "cudaGraphicsUnmapResources" << std::endl;
    	// Unmap the resource
    	checkCudaErrors( cudaGraphicsUnmapResources(1, &res_data));
    }

The problem is when executing the program CUDA throws an error:

> CUDA error at
> e:\workspace\master\gradientrenderer\gradientrenderer\source.cpp:175
> code=11(cudaErrorInvalidValue) "cudaMemcpy2DFromArray(dData,
> window_w*4*sizeof(float), cuda_data ,0 ,0, window_w*4*sizeof(float),
> window_h, cudaMemcpyDeviceToDevice)"

Can someone tell me what I am doing wrong? As this is part of a bigger project using a predefined kernel (no way to change this kernel) I need to call cudaMemcpy2DFromArray so that the texture is correctly transferred to the kernel.

1 Like