I just tested it:

1.18:

    BenchmarkMap-10         13388006                77.38 ns/op
    BenchmarkSwitch-10      50482033                23.00 ns/op
    BenchmarkSlice-10       100000000               10.90 ns/op

1.19:

    BenchmarkMap-10         14947908                76.78 ns/op
    BenchmarkSwitch-10      49444435                23.01 ns/op
    BenchmarkSlice-10       100000000               10.74 ns/op

edit: If I understand the release notes correctly the switch optimisations are only new for large integers and strings.