如何使用标准库测试Go函数的执行时间

Question

I've written the following function that when used with Defer will surely do the trick but is there something in standard library I could use for this? I'm looking something similar to Python timeit which I could use directly from the shell?

package main

import (
    "fmt"
    "time"
)

func main() {
    defer timeTrack(time.Now(), "looptest")
    nums := []int{1, 2, 3, 4, 5, 6, 7, 8, 9, 10}
    var m []int
    for n := range nums {
        if n%2 == 0 {
            m = append(m, n)
        }
    }
}

func timeTrack(start time.Time, name string) {
    elapsed := time.Since(start)
    fmt.Printf("%s took %s", name, elapsed)
}

Answer 1

Use the Go testing package. For example,

main.go:

package main

func doStuff() {
    nums := []int{1, 2, 3, 4, 5, 6, 7, 8, 9, 10}
    var m []int
    for n := range nums {
        if n%2 == 0 {
            m = append(m, n)
        }
    }
}

func main() {
    doStuff()
}

main_test.go:

package main

import "testing"

func BenchmarkStuff(b *testing.B) {
    for i := 0; i < b.N; i++ {
        doStuff()
    }
}

Output:

$ go test -bench=. -benchmem
goos: linux
goarch: amd64
pkg: main
BenchmarkStuff-4    5000000    276 ns/op    120 B/op    4 allocs/op

---

Using the information from the benchmark, you can improve performance. Reduce the number and size of allocations by using an estimate to initialize slice m.

func doStuff() {
    nums := []int{1, 2, 3, 4, 5, 6, 7, 8, 9, 10}
    m := make([]int, 0, len(nums)/2+1)
    for n := range nums {
        if n%2 == 0 {
            m = append(m, n)
        }
    }
}

Output:

$ go test -bench=. -benchmem
goos: linux
goarch: amd64
pkg: main
BenchmarkStuff-4    20000000    83.1 ns/op    48 B/op    1 allocs/op