issue: benchmark of reverse_integer

[WindomZ]

Scene

Run the script:

git clone -b issues https://github.com/WindomZ/leetcode.go "$YOUR_PROJECT_PATH"
cd "$YOUR_PROJECT_PATH"
go test -test.bench=".*" ./issues/reverse_integer

Question

1 number of CPUs:

Benchmark_A_reverse_1   100000000               15.0 ns/op
Benchmark_A_reverse_2   100000000               14.9 ns/op
Benchmark_A_reverse_3   100000000               14.2 ns/op
Benchmark_A_reverse_4   100000000               14.1 ns/op
Benchmark_B_reverse_3   100000000               14.1 ns/op
Benchmark_B_reverse_4   100000000               14.1 ns/op
Benchmark_B_reverse_1   100000000               14.9 ns/op
Benchmark_B_reverse_2   100000000               14.9 ns/op
Benchmark_C_reverse_1   100000000               14.9 ns/op
Benchmark_C_reverse_3   100000000               14.1 ns/op
Benchmark_C_reverse_2   100000000               15.0 ns/op
Benchmark_C_reverse_4   100000000               14.2 ns/op

8 number of CPUs:

Benchmark_A_reverse_1-8         100000000               14.9 ns/op
Benchmark_A_reverse_2-8         100000000               14.9 ns/op
Benchmark_A_reverse_3-8         100000000               14.2 ns/op
Benchmark_A_reverse_4-8         100000000               14.1 ns/op
Benchmark_B_reverse_3-8         100000000               14.1 ns/op
Benchmark_B_reverse_4-8         100000000               14.1 ns/op
Benchmark_B_reverse_1-8         100000000               14.9 ns/op
Benchmark_B_reverse_2-8         100000000               14.9 ns/op
Benchmark_C_reverse_1-8         100000000               14.9 ns/op
Benchmark_C_reverse_3-8         100000000               14.2 ns/op
Benchmark_C_reverse_2-8         100000000               15.0 ns/op
Benchmark_C_reverse_4-8         100000000               14.4 ns/op

Why reverse_3 and reverse_4 are faster than reverse_1 and reverse_2?