Ring Topology

Appears in 1 paper

An arrangement of P GPUs in a logical circle where GPU i communicates with GPU i-1 (receives data) and GPU i+1 (sends data).

As used in Paper 19 — Ring Attention with Blockwise Transformers for Near-Infinite Context →

An arrangement of P GPUs in a logical circle where GPU i communicates with GPU i-1 (receives data) and GPU i+1 (sends data). Each device has only two neighbours. Avoids all-to-all communication, making the approach scalable.