This benchmark reads in file <tempdir>/gopacket_benchmark.pcap and measures the time it takes to decode all packets from that file. If the file doesn't exist, it's pulled down from a publicly available location. However, you can feel free to substitute your own file at that location, in which case the benchmark will run on your own data.
It's also useful for figuring out which packets may be causing errors. Pass in the --printErrors flag, and it'll print out error layers for each packet that has them. This includes any packets that it's just unable to decode, which is a great way to find new protocols to decode, and get test packets to write tests for them.