[yeoyeo42]

5 microseconds is a whole lot of time for just a branch misprediction, which is on the order of a few nanoseconds. probably a lot more important stuff going on and focusing on the wrong thing.

for example, predicting a different branch can lead to different memory access patterns. maybe you can get rid of most of the real world cost of the misprediction by just forcefully prefetching the memory the send() path is going to use.