cudaGraphicsGLRegisterImage for RenderBuffer

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi all,

I am trying to map an opengl render buffer to a cuda pointer by using cudaGraphicsGLRegisterImage, but the function returns the error cudaErrorInvalidValue.

The render buffer is associated to the deth buffer of a frame buffer.

glGenFramebuffersEXT(1, &_fbo);

	glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, _fbo);

	glGenRenderbuffersEXT(1, &_rbo);

	glBindRenderbufferEXT(GL_RENDERBUFFER_EXT, _rbo);

	glRenderbufferStorageEXT(GL_RENDERBUFFER_EXT, GL_DEPTH_COMPONENT, width(), height());

	glBindRenderbufferEXT(GL_RENDERBUFFER_EXT, 0);

	// attach a texture to FBO color attachement point

	glFramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, GL_COLOR_ATTACHMENT0_EXT, GL_TEXTURE_2D, _tex, 0);

	// attach a renderbuffer to depth attachment point

	glFramebufferRenderbufferEXT(GL_FRAMEBUFFER_EXT, GL_DEPTH_ATTACHMENT_EXT, GL_RENDERBUFFER_EXT, _rbo);

	glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, 0);

	cutilSafeCall( cudaGraphicsGLRegisterImage(&_cuda_resources[1], _rbo, GL_RENDERBUFFER, cudaGraphicsMapFlagsNone ) );  ==> cudaErrorInvalidValue

When I do it with a 2D texture it works well

cutilSafeCall( cudaGraphicsGLRegisterImage(&_cuda_resources[1], _tex, GL_GL_TEXTURE_2D, cudaGraphicsMapFlagsNone ) );

Do I have to respect something regarding the render buffer?

edit: do you know if it is possible to do the same with a 3D texture?

Thanks

–pium