While doing some benchmarking, I noticed that strings.IndexByte was slower than a simple index-based for loop. This was originally reported here:
https://groups.google.com/forum/#!topic/golang-nuts/Vjy_UUcugPE

Keith observed that the benchmark ran faster when the string length was increased to 16 characters.

For a < 16 character string, the results change dramatically:

BenchmarkGetHostIndexByte-8     100000000           23.3 ns/op
BenchmarkGetHostIndexLoop-8     200000000            9.92 ns/op

For a 16 character string, the benchmarks are:

BenchmarkGetHostIndexByte-8     200000000            5.91 ns/op
BenchmarkGetHostIndexLoop-8     200000000            9.17 ns/op

Keith mentioned that the REP instruction was most likely to blame.