if ip cef isn’t enabled by default which I think it will be then you can enable it
sh ip cef
io cef enable
ip load-sharing per-packet
Please rate and mark as an accepted solution if you have found any of the information provided useful. This then could assist others on these forums to find a valuable answer and broadens the community’s global network.
The original poster asked a specific question about how to enable per packet load balancing. And Paul has provided an appropriate answer. I would ask the original poster if they are sure that they want to do this. Per packet sounds like it might be a good thing, and the attempt to equalize the load over connections sounds like it would be a good thing. But per packet load balancing is likely to result in some out of order packets. Depending on the applications involved this might be trivial or it might be significant. I have had the experience with several customers of turning on per packet load balancing and seeing the performance of critical applications get worse.
As Rick notes, many applications (especially some of the real-time applications) do not like out-of-order packets, which per-packet load sharing will often cause. Even an ordinary TCP data transfer, e.g. FTP, can be impacted as some TCP stacks will begin a re-transmission when it believes a packet is lost, yet it's not, if it's only just delayed because of being out-of-order.
If you're working with "slow" WAN links, something like MLPPP might be used, which will well load balance and not re-order packet sequences. If you're working with "fast" links, you might look into use PfR which can dynamically load balance L3 links.
I believe (?) later platforms now usually have CEF on by default and recall (?) that some recent platforms won't allow you to disable CEF (as it's so critical to so many of Cisco's later IOS features).
That said, what Paul lists will show the current status of CEF (i.e. whether it's on or not), how to enable it if not and finally, Paul shows how to enable per-packet load sharing on an interface.