An 8x8 2D torus network topology connects 64 nodes in a grid where each node links horizontally and vertically to neighbors, with wrap-around edges creating a closed loop. Horizontal links connect adjacent nodes along rows (0,0)→(1,0)→...→(7,0), while vertical links connect columns (0,0)→(0,1)→...→(0,7), and wrap-around links close the grid by connecting edge nodes back to opposite edges. This topology eliminates dead ends and reduces network diameter, making it ideal for supercomputing clusters, distributed memory systems, and low-latency parallel computing where uniform hop counts and fault tolerance matter. Fork this diagram to customize node counts, add latency annotations, or adapt it for your HPC or interconnect documentation. Torus networks are foundational in systems like Cray supercomputers and modern AI training clusters requiring predictable, scalable communication patterns.