AcceptToMemoryPool is the primary bottleneck so I wonder what the real capacity of BCH network is. We can then calculate the maximum capacity based on ATMP bottleneck and then stress