问题描述

在本机测试用例中,初始化键值对100W,初始化容量200W(其实当键值对更少,比如10W,5W等级别,情况也是类似的),发现初始化容量的效率和不初始化容量的效率并不是稳定不变的,为何一直听说当可预估map内容大小时,应该预填容量值呢?

测试代码

int puts = 1000000, cap = puts * 2; for (int k = 0; k < 5; k++) { System.out.println("=========独立测试开始=========="); HashMap map5 = new HashMap(cap); long time01Start = System.currentTimeMillis(); for (int i = 0; i < puts; i++) { map5.put(i, i); } System.out.println("HashMap初始化容量" + puts / 10000 + "W的耗时:" + (System.currentTimeMillis() - time01Start) + "ms"); HashMap map6 = new HashMap(); long time02Start = System.currentTimeMillis(); for (int i = 0; i < puts; i++) { map6.put(i, i); } System.out.println("HashMap未初始化容量" + puts / 10000 + "W的耗时:" + (System.currentTimeMillis() - time02Start) + "ms"); System.out.println("=========独立测试结束=========="); }

5次的测试结果

=========独立测试开始========== HashMap初始化容量100W的耗时:964ms HashMap未初始化容量100W的耗时:603ms =========独立测试结束========== =========独立测试开始========== HashMap初始化容量100W的耗时:119ms HashMap未初始化容量100W的耗时:658ms =========独立测试结束========== =========独立测试开始========== HashMap初始化容量100W的耗时:26ms HashMap未初始化容量100W的耗时:72ms =========独立测试结束========== =========独立测试开始========== HashMap初始化容量100W的耗时:154ms HashMap未初始化容量100W的耗时:34ms =========独立测试结束========== =========独立测试开始========== HashMap初始化容量100W的耗时:134ms HashMap未初始化容量100W的耗时:34ms =========独立测试结束==========