一千万个随机数排序,如何24秒蜕变成3秒？如何从700M内存消耗变成200M？

上一篇文章写的十分的烂,经过科普看语言源码实现用的是quicksort实现的底层排序,在这里模仿一下,勿喷！

package main

import (
	"fmt"
	"math/rand"
	"runtime"
	"sort"
	"time"
)

func mergeonce(l, r []int) []int {
	m := make([]int, 0, len(l)+len(r))
	i, j := 0, 0
	if i < len(l) && j < len(r) {
		for {
			if l[i] < l[j] {
				m = append(m, l[i])
				i++
				if i == len(l) {
					break
				}
			} else {
				m = append(m, l[j])
				j++
				if j == len(r) {
					break
				}
			}
		}
	}
	m = append(append(m, l[i:]...), r[j:]...)
	return m
}

func merge(in chan []int, out chan []int) {
	var next chan []int
	var l []int
	for list := range in {
		if l == nil {
			l = list
			continue
		}

		r := list

		if next == nil {
			next = make(chan []int, 1)
			go merge(next, out)
		}

		next <- mergeonce(l, r)
		l = nil
	}
	if next == nil {
		out <- l
	} else {
		if l != nil {
			next <- l
		}
		close(next)
	}
}

func main() {
	runtime.GOMAXPROCS(2)
	ch := make(chan []int, 1)

	const Num = 100
	const WNum = 100
	fmt.Println(time.Now())
	go func(n int, out chan []int) {
		for i := 0; i < n; i++ {
			list := make([]int, 1000)
			for j := range list {
				list[j] = rand.Int()
			}

			sort.Ints(list)
			out <- list
		}
		close(out)
	}(Num*WNum, ch)

	out := make(chan []int)
	go merge(ch, out)
	list := <-out
	fmt.Println(time.Now())
	fmt.Println(len(list))
	fmt.Println(sort.IntsAreSorted(list))
}

一千万个随机数排序,如何24秒蜕变成3秒？如何从700M内存消耗变成200M？

原文：http://blog.csdn.net/fyxichen/article/details/46572003