A binaural rendering addon based on the EBU ADM Renderer has been developed, which renders using virtual loudspeaker systems. During development, two optimization approaches emerged regarding rendering quality and computational costs. The first approach deals with the rendering of coherent signals from different positions, which is particularly relevant for virtual speakers since object positioning is based on amplitude panning. The second approach, on the other hand, deals with the reduction of computational costs when rendering the binaural room response by using different virtual speaker setups for the rendering of the direct path and the room response. To evaluate both approaches, a listening test was conducted. The results of this test showed, that both approaches positively influenced rendering quality or performance.