Windows Server NIC Teaming and bandwidth aggregation - what is the real expectation?

There seems to be some difference of opinion on the effect of NIC Teaming on bandwidth aggregation.

From my tests using dual 1GB teaming, I see up to 2GB throughput in or out, as long as that data is going to or from more than one other servers (i.e. 1GB each to or from two other servers). Server-to-server I only see 1GB maximum bandwidth at any point, regardless of how many 1GB NICs are in their teams. This seems to match the conclusion I've come to from most of what I have read.

But then I come across the odd reference where it seems that bandwidth is aggregated as if it was the native speed of a single connection.

For example, in the last post on https://serverfault.com/questions/754502/windows-2012-r2-nic-teaming-does-not-aggregate-bandwidth, JJ Nace seems to imply that using Switch Independent mode with Dynamic load balancing does aggregate bandwidth in every sense in their setup.

Could someone provide a definitive answer (ideally with source) on whether bandwidth can be aggregated as if it were a single connection? If it is possible, what configuration is necessary?

Thanks.