Change io_tx_batch() function to use DPDK's packet reordering library (optionally) - http://dpdk.org/browse/dpdk/tree/examples/packet_ordering/main.c
This may impact the latency distribution a lot, especially for adaptive load balancing configurations where we mix CPU and GPU.
Change
io_tx_batch()function to use DPDK's packet reordering library (optionally) - http://dpdk.org/browse/dpdk/tree/examples/packet_ordering/main.c This may impact the latency distribution a lot, especially for adaptive load balancing configurations where we mix CPU and GPU.