The overall idea of wormhole flow control is to have the packets "inch" between nodes like a worm would inch along a branch. Once the head of the "worm" moves, the rest of the "worm" will follow along the same path one flit at a time. This makes the request travel much faster through the nodes, and frees up any resources left after the tail of the worm has passed through.
This comment was marked helpful 0 times.
Because flits are much smaller than regular packets, it naturally follows that a network using wormhole flow control requires much less buffer space within its nodes than store-and-forward or cut-through based networks. This could lead to reduced hardware costs.