Is unity uses rendering from the cpu or gpu ?

The GPU does the hard work. All objects that are going to be rendered are ultimately meshes composed by lots of triangles and some textures used to paint them (and to add some details, like bump maps, occlusion maps etc.). Unity (CPU) selects w$$anonymous$$ch objects are candidates to be rendered, then sends them one by one to the GPU: for each object, Unity sets up the shaders and textures and starts feeding the GPU with the vertices, triangles, normals and other attributes associated to the vertices. The actual rendering is done by the GPU, where the shader calculates the coordinates of each visible triangle on the screen (vertex section) and paints the triangle according to the textures and lighting (fragment section). T$$anonymous$$s process is repeated every frame for each active camera, and then the resulting buffer (where stuff actually is rendered) is finally transferred to the screen (kind of: some cameras may render to other buffers in order to create effects like reflections, ambient lighting, post processing etc.)

Both/Either. You can create meshes in code on the CPU, or in a geometry shader on the GPU. You can batch instances on the CPU, or using instancing on the GPU. You can perform calculations on the CPU or in a compute shader on the GPU....