This paper presents a high-performance and low-power hardware architecture for real-time sound rendering on mobile devices. Traditional sound rendering algorithms require high-performance CPUs or GPUs for processing because of its high computational complexities to realize ultra-realistic 3D audio. Thus, it has been hard to achieve real-time rates on low-power mobile devices. To overcome this limitation, we propose a hardware architecture that adopts hardware-friendly sound-propagation-path calculation algorithms. We verified the function and performance of our architecture through its implementation on an FPGA board. According to ASIC evaluation with the 8-nm process technology, it achieves high performance with 120 FPS, low power consumption with 50 mW, and a small silicon area with 0.31 mm2, allowing real-time sound rendering on mobile devices.