Hi, I tested the performance of "test-pipeline"(--none) and "l2fwd", but the result confused me. "l2fwd" was three times better than "test-pipeline" in pps. I set them to the same parameters like this: "l2fwd": MAX_PKT_BURST=64 nb_rxd = 1024 nb_txd = 1024 "test-pipeline" burst_size_rx_read = 64 burst_size_rx_write = 64 burst_size_worker_read = 64 burst_size_worker_write = 64 burst_size_tx_read = 64 burst_size_tx_write = 64 port_rx_ring_size = 1024 port_tx_ring_size = 1024 ring_rx_size = 1024 ring_tx_size = 1024 Can anyone know this reason and tell me? Thank you in advance. Best wishes!