Ethernet Switch Frame Latency: Store-and-Forward vs Cut-Through (2D)

Every switched Ethernet frame's real-world latency is the sum of two independent physical delays: propagation delay, set by cable length and the ~0.65c signal velocity in twisted-pair copper, and serialization delay, set by the link's bit rate and the frame's size. This 2D simulator computes both from scratch for six hosts sitting at their own real distances from a central switch, and lets you compare store-and-forward switching (which must fully buffer a frame before relaying it) against cut-through switching (which starts relaying as soon as the destination MAC address has arrived). The switch still runs the real MAC-learning algorithm underneath — flooding an unknown destination to every other port, forwarding unicast once a host's port is known, and aging stale entries out of the table — but now every packet's travel time on screen is driven by the actual computed microsecond latency for its own link, so a distant host in a flood visibly arrives later than a near one, and a 10 Mbps link visibly crawls compared to 1000 Mbps.