Increasing ndis filter driver speed with deferred procedure calls on target processors

knowing more about what kind of analysis you are trying to do will help, but I suggest you start by grouping TCP packets by connection (protocol, source IP, source port, destination IP & destination port) and routing each of these packet flows onto a single core. You might need to ensure that the reverse traffic is routed via the same core depending on the analysis that you are doing

further optimizations usually require knowledge of the later 4 and higher protocols of interest