Home > Java > Java Tutorial > body text

Summary of 35 Java code performance optimizations

伊谢尔伦
Release: 2016-11-30 11:46:16
Original
1266 people have browsed it

 Foreword

 Code optimization is a very important topic. Some people may think it is useless. What are the small things that can be modified? What impact will the modification or not have on the running efficiency of the code? I think about this question like this, just like a whale in the sea, is it useful for it to eat a small shrimp? It was useless, but after eating more shrimps, the whale was full. The same is true for code optimization. If the project focuses on launching without bugs as soon as possible, then you can focus on the big and let go of the small at this time, and the details of the code do not need to be refined; but if there is enough time to develop and maintain the code, you must consider every aspect at this time. There are details that can be optimized. The accumulation of small optimization points one by one will definitely improve the running efficiency of the code.

 The goals of code optimization are:

 1. Reduce the size of the code

 2. Improve the efficiency of code running

Code optimization details

 1. Try to specify the final modifier of classes and methods

 with final modification The symbol class is not derived. In the Java core API, there are many examples of final applications, such as java.lang.String, where the entire class is final. Specifying the final modifier for a class prevents the class from being inherited, and specifying the final modifier for a method prevents the method from being overridden. If a class is designated as final, all methods of the class are final. The Java compiler will look for opportunities to inline all final methods. Inlining plays a significant role in improving Java running efficiency. For details, see Java runtime optimization. This can improve performance by an average of 50%.

  2. Try to reuse objects

Especially when using String objects, StringBuilder/StringBuffer should be used instead when string connections occur. Since the Java virtual machine not only spends time generating objects, it may also need to spend time garbage collecting and processing these objects in the future. Therefore, generating too many objects will have a great impact on the performance of the program.

 3. Use local variables as much as possible

 The parameters passed when calling the method and the temporary variables created during the call are saved in the stack for faster speed. Other variables, such as static variables, instance variables, etc., are created in the heap , slower. In addition, the contents of variables created in the stack are gone as the method ends, and no additional garbage collection is required.

 4. Close the stream in a timely manner

 During the Java programming process, be careful when performing database connections and I/O stream operations. After use, close the stream in time to release resources. Because operations on these large objects will cause a lot of system overhead, a little carelessness will lead to serious consequences.

 5. Minimize repeated calculations of variables

 Clear a concept. Calling a method, even if there is only one statement in the method, is costly, including creating a stack frame, protecting the scene when calling the method, and restoring when the method is called. Waiting on site. So for example, the following operation:

for (int i = 0; i < list.size(); i++)
{...}
Copy after login

 It is recommended to replace it with:

for (int i = 0, int length = list.size(); i < length; i++)
{...}
Copy after login

 In this way, when list.size() is very large, it will reduce a lot of consumption

 6、尽量采用懒加载的策略,即在需要的时候才创建

  例如:

String str = "aaa";if (i == 1)
{
list.add(str);
}
Copy after login

  建议替换为:

if (i == 1)
{
String str = "aaa";
list.add(str);
}
Copy after login

  7、慎用异常

  异常对性能不利。抛出异常首先要创建一个新的对象,Throwable接口的构造函数调用名为fillInStackTrace()的本地同步方法,fillInStackTrace()方法检查堆栈,收集调用跟踪信息。只要有异常被抛出,Java虚拟机就必须调整调用堆栈,因为在处理过程中创建了一个新的对象。异常只能用于错误处理,不应该用来控制程序流程。

  8、不要在循环中使用try…catch…,应该把其放在最外层

  除非不得已。如果毫无理由地这么写了,只要你的领导资深一点、有强迫症一点,八成就要骂你为什么写出这种垃圾代码来了

 9、如果能估计到待添加的内容长度,为底层以数组方式实现的集合、工具类指定初始长度

  比如ArrayList、LinkedLlist、StringBuilder、StringBuffer、HashMap、HashSet等等,以StringBuilder为例:

  (1)StringBuilder()      // 默认分配16个字符的空间

  (2)StringBuilder(int size)  // 默认分配size个字符的空间

  (3)StringBuilder(String str) // 默认分配16个字符+str.length()个字符空间

  可以通过类(这里指的不仅仅是上面的StringBuilder)的来设定它的初始化容量,这样可以明显地提升性能。比如StringBuilder吧,length表示当前的StringBuilder能保持的字符数量。因为当StringBuilder达到最大容量的时候,它会将自身容量增加到当前的2倍再加2,无论何时只要StringBuilder达到它的最大容量,它就不得不创建一个新的字符数组然后将旧的字符数组内容拷贝到新字符数组中—-这是十分耗费性能的一个操作。试想,如果能预估到字符数组中大概要存放5000个字符而不指定长度,最接近5000的2次幂是4096,每次扩容加的2不管,那么:

  (1)在4096 的基础上,再申请8194个大小的字符数组,加起来相当于一次申请了12290个大小的字符数组,如果一开始能指定5000个大小的字符数组,就节省了一倍以上的空间

  (2)把原来的4096个字符拷贝到新的的字符数组中去

  这样,既浪费内存空间又降低代码运行效率。所以,给底层以数组实现的集合、工具类设置一个合理的初始化容量是错不了的,这会带来立竿见影的效果。但是,注意,像HashMap这种是以数组+链表实现的集合,别把初始大小和你估计的大小设置得一样,因为一个table上只连接一个对象的可能性几乎为0。初始大小建议设置为2的N次幂,如果能估计到有2000个元素,设置成new HashMap(128)、new HashMap(256)都可以。

  10、当复制大量数据时,使用System.arraycopy()命令

  11、乘法和除法使用移位操作

  例如:

for (val = 0; val < 100000; val += 5)
{
a = val * 8;
b = val / 2;
}
Copy after login

  用移位操作可以极大地提高性能,因为在计算机底层,对位的操作是最方便、最快的,因此建议修改为:

for (val = 0; val < 100000; val += 5)
{
a = val << 3;
b = val >> 1;
}
Copy after login

  移位操作虽然快,但是可能会使代码不太好理解,因此最好加上相应的注释。

 12、循环内不要不断创建对象引用

  例如:

for (int i = 1; i <= count; i++)
{
Object obj = new Object();
}
Copy after login

  这种做法会导致内存中有count份Object对象引用存在,count很大的话,就耗费内存了,建议为改为:

Object obj = null;for (int i = 0; i <= count; i++) { obj = new Object(); }

  这样的话,内存中只有一份Object对象引用,每次new Object()的时候,Object对象引用指向不同的Object罢了,但是内存中只有一份,这样就大大节省了内存空间了。

  13、基于效率和类型检查的考虑,应该尽可能使用array,无法确定数组大小时才使用ArrayList

  14、尽量使用HashMap、ArrayList、StringBuilder,除非线程安全需要,否则不推荐使用Hashtable、Vector、StringBuffer,后三者由于使用同步机制而导致了性能开销

 15、不要将数组声明为public static final

  因为这毫无意义,这样只是定义了引用为static final,数组的内容还是可以随意改变的,将数组声明为public更是一个安全漏洞,这意味着这个数组可以被外部类所改变

  16、尽量在合适的场合使用单例

  使用单例可以减轻加载的负担、缩短加载的时间、提高加载的效率,但并不是所有地方都适用于单例,简单来说,单例主要适用于以下三个方面:

  (1)控制资源的使用,通过线程同步来控制资源的并发访问

  (2)控制实例的产生,以达到节约资源的目的

  (3)控制数据的共享,在不建立直接关联的条件下,让多个不相关的进程或线程之间实现通信

  17、尽量避免随意使用静态变量

  要知道,当某个对象被定义为static的变量所引用,那么gc通常是不会回收这个对象所占有的堆内存的,如:

public class A
{ 
private static B b = new B();
}
Copy after login

  此时静态变量b的生命周期与A类相同,如果A类不被卸载,那么引用B指向的B对象会常驻内存,直到程序终止

 18、及时清除不再需要的会话

  为了清除不再活动的会话,许多应用服务器都有默认的会话超时时间,一般为30分钟。当应用服务器需要保存更多的会话时,如果内存不足,那么操作系统会把部分数据转移到磁盘,应用服务器也可能根据MRU(最近最频繁使用)算法把部分不活跃的会话转储到磁盘,甚至可能抛出内存不足的异常。如果会话要被转储到磁盘,那么必须要先被序列化,在大规模集群中,对对象进行序列化的代价是很昂贵的。因此,当会话不再需要时,应当及时调用HttpSession的invalidate()方法清除会话。

  19、实现RandomAccess接口的集合比如ArrayList,应当使用最普通的for循环而不是foreach循环来遍历

  这是JDK推荐给用户的。JDK API对于RandomAccess接口的解释是:实现RandomAccess接口用来表明其支持快速随机访问,此接口的主要目的是允许一般的算法更改其行为,从而将其应用到随机或连续访问列表时能提供良好的性能。实际经验表明,实现RandomAccess接口的类实例,假如是随机访问的,使用普通for循环效率将高于使用foreach循环;反过来,如果是顺序访问的,则使用Iterator会效率更高。可以使用类似如下的代码作判断:

if (list instanceof RandomAccess)
{ for (int i = 0; i < list.size(); i++){}
}else{
Iterator iterator = list.iterable(); while (iterator.hasNext()){iterator.next()}
}
Copy after login

  foreach循环的底层实现原理就是迭代器Iterator,参见Java语法糖1:可变长度参数以及foreach循环原理。所以后半句”反过来,如果是顺序访问的,则使用Iterator会效率更高”的意思就是顺序访问的那些类实例,使用foreach循环去遍历。

 20. Use synchronized code blocks instead of synchronized methods

This has been clearly stated in the article synchronized lock method blocks in multi-threaded modules. Unless it is certain that an entire method needs to be synchronized, try to use synchronization as much as possible. Code blocks avoid synchronizing code that does not need to be synchronized, which affects code execution efficiency.

 21. Declare the constants as static final and name them in uppercase letters

 This way you can put these contents into the constant pool during compilation to avoid calculating the value of the constant during runtime. In addition, naming constants in uppercase letters can also make it easier to distinguish constants from variables

 22. Do not create some unused objects and do not import some unused classes

 This is meaningless, if "The value" appears in the code of the local variable i is not used”, “The import java.util is never used”, then please delete these useless contents

 23. Avoid using reflection during program running

 About, please see reflection. Reflection is a very powerful function provided by Java to users. Powerful functions often mean low efficiency. It is not recommended to use the reflection mechanism during the running of the program, especially the frequent use of the reflection mechanism, especially the method's invoke method. If it is really necessary, a recommended approach is to use reflection instances for those classes that need to be loaded through reflection when the project is started. Create an object and put it into memory - the user only cares about getting the fastest response when interacting with the peer, and does not care about how long it takes for the peer's project to start.

 24. Use database connection pool and thread pool

 These two pools are used to reuse objects. The former can avoid frequently opening and closing connections, and the latter can avoid frequently creating and destroying threads

 25. Use Buffered input and output streams perform IO operations

Buffered input and output streams, namely BufferedReader, BufferedWriter, BufferedInputStream, BufferedOutputStream, which can greatly improve IO efficiency

 26. Use ArrayList in scenarios with a lot of sequential insertion and random access. LinkedList is used in scenarios where there are many element deletions and intermediate insertions

 This is known by understanding the principles of ArrayList and LinkedList

 27. Don’t have too many formal parameters in the public method

 The public method is the method provided to the outside world. If Giving these methods too many formal parameters has two main disadvantages:

 1. It violates the object-oriented programming idea. Java emphasizes that everything is an object. Too many formal parameters are not consistent with the object-oriented programming idea

2. Too many parameters will inevitably lead to an increase in the error probability of method calls

As for how many parameters this "too many" refers to, let's say 3 or 4. For example, we use JDBC to write an insertStudentInfo method. There are 10 student information fields to be inserted into the Student table. These 10 parameters can be encapsulated in an entity class as the formal parameters of the insert method

  28、字符串变量和字符串常量equals的时候将字符串常量写在前面

  这是一个比较常见的小技巧了,如果有以下代码:

String str = "123";
if (str.equals("123")) {
...
}
Copy after login

  建议修改为:

String str = "123";
if ("123".equals(str))
{
...
}
Copy after login

  这么做主要是可以避免空指针异常

 29、请知道,在java中if (i == 1)和if (1 == i)是没有区别的,但从阅读习惯上讲,建议使用前者

  平时有人问,”if (i == 1)”和”if (1== i)”有没有区别,这就要从C/C++讲起。

  在C/C++中,”if (i == 1)”判断条件成立,是以0与非0为基准的,0表示false,非0表示true,如果有这么一段代码:

int i = 2;
if (i == 1)
{
...
}else{
...
}
Copy after login

  C/C++判断”i==1″不成立,所以以0表示,即false。但是如果:

int i = 2;if (i = 1) { ... }else{ ... }
Copy after login

  万一程序员一个不小心,把”if (i == 1)”写成”if (i = 1)”,这样就有问题了。在if之内将i赋值为1,if判断里面的内容非0,返回的就是true了,但是明明i为2,比较的值是1,应该返回的false。这种情况在C/C++的开发中是很可能发生的并且会导致一些难以理解的错误产生,所以,为了避免开发者在if语句中不正确的赋值操作,建议将if语句写为:

int i = 2;if (1 == i) { ... }else{ ... }
Copy after login

  这样,即使开发者不小心写成了”1 = i”,C/C++编译器也可以第一时间检查出来,因为我们可以对一个变量赋值i为1,但是不能对一个常量赋值1为i。

  但是,在Java中,C/C++这种”if (i = 1)”的语法是不可能出现的,因为一旦写了这种语法,Java就会编译报错”Type mismatch: cannot convert from int to boolean”。但是,尽管Java的”if (i == 1)”和”if (1 == i)”在语义上没有任何区别,但是从阅读习惯上讲,建议使用前者会更好些。

  30、不要对数组使用toString()方法

  看一下对数组使用toString()打印出来的是什么:

public static void main(String[] args)
{ int[] is = new int[]{1, 2, 3};
System.out.println(is.toString());
}
Copy after login

  结果是:

[I@18a992f
Copy after login

  本意是想打印出数组内容,却有可能因为数组引用is为空而导致空指针异常。不过虽然对数组toString()没有意义,但是对集合toString()是可以打印出集合里面的内容的,因为集合的父类AbstractCollections重写了Object的toString()方法。

 31、不要对超出范围的基本数据类型做向下强制转型

  这绝不会得到想要的结果:

public static void main(String[] args)
{ 
long l = 12345678901234L;
int i = (int)l;
System.out.println(i);
}
Copy after login

  我们可能期望得到其中的某几位,但是结果却是:

  1942892530

  解释一下。Java中long是8个字节64位的,所以12345678901234在计算机中的表示应该是:

  0000 0000 0000 0000 0000 1011 0011 1010 0111 0011 1100 1110 0010 1111 1111 0010

  一个int型数据是4个字节32位的,从低位取出上面这串二进制数据的前32位是:

  0111 0011 1100 1110 0010 1111 1111 0010

  这串二进制表示为十进制1942892530,所以就是我们上面的控制台上输出的内容。从这个例子上还能顺便得到两个结论:

  1、整型默认的数据类型是int,long l = 12345678901234L,这个数字已经超出了int的范围了,所以最后有一个L,表示这是一个long型数。顺便,浮点型的默认类型是double,所以定义float的时候要写成”"float f = 3.5f”

  2、接下来再写一句”int ii = l + i;”会报错,因为long + int是一个long,不能赋值给int

  32、公用的集合类中不使用的数据一定要及时remove掉

  如果一个集合类是公用的(也就是说不是方法里面的属性),那么这个集合里面的元素是不会自动释放的,因为始终有引用指向它们。所以,如果公用集合里面的某些数据不使用而不去remove掉它们,那么将会造成这个公用集合不断增大,使得系统有内存泄露的隐患。

 33、把一个基本数据类型转为字符串,基本数据类型.toString()是最快的方式、String.valueOf(数据)次之、数据+”"最慢

  把一个基本数据类型转为一般有三种方式,我有一个Integer型数据i,可以使用i.toString()、String.valueOf(i)、i+”"三种方式,三种方式的效率如何,看一个测试:

public static void main(String[] args)
{ 
int loopTime = 50000;
Integer i = 0; long startTime = System.currentTimeMillis(); for (int j = 0; j < loopTime; j++)
{
String str = String.valueOf(i);
}
System.out.println("String.valueOf():" + (System.currentTimeMillis() - startTime) + "ms");
startTime = System.currentTimeMillis(); for (int j = 0; j < loopTime; j++)
{
String str = i.toString();
}
System.out.println("Integer.toString():" + (System.currentTimeMillis() - startTime) + "ms");
startTime = System.currentTimeMillis(); for (int j = 0; j < loopTime; j++)
{
String str = i + "";
}
System.out.println("i + \"\":" + (System.currentTimeMillis() - startTime) + "ms");
}
Copy after login

  运行结果为:

String.valueOf():11ms Integer.toString():5ms i + "":25ms

  所以以后遇到把一个基本数据类型转为String的时候,优先考虑使用toString()方法。至于为什么,很简单:

  1、String.valueOf()方法底层调用了Integer.toString()方法,但是会在调用前做空判断

  2、Integer.toString()方法就不说了,直接调用了

  3、i + “”底层使用了StringBuilder实现,先用append方法拼接,再用toString()方法获取字符串

  三者对比下来,明显是2最快、1次之、3最慢

 34、使用最有效率的方式去遍历Map

  遍历Map的方式有很多,通常场景下我们需要的是遍历Map中的Key和Value,那么推荐使用的、效率最高的方式是:

public static void main(String[] args)
{
HashMap hm = new HashMap();
hm.put("111", "222");
Set> entrySet = hm.entrySet();
Iterator> iter = entrySet.iterator(); while (iter.hasNext())
{
Map.Entry entry = iter.next();
System.out.println(entry.getKey() + "\t" + entry.getValue());
}
}
Copy after login

  如果你只是想遍历一下这个Map的key值,那用”Set keySet = hm.keySet();”会比较合适一些

 35、对资源的close()建议分开操作

  意思是,比如我有这么一段代码:

try{
XXX.close();
YYY.close();
}catch (Exception e)
{
...
}
Copy after login

  建议修改为:

try{ XXX.close(); }catch (Exception e) { ... }try{ YYY.close(); }catch (Exception e) { ... }
Copy after login

  虽然有些麻烦,却能避免资源泄露。我们想,如果没有修改过的代码,万一XXX.close()抛异常了,那么就进入了cath块中了,YYY.close()不会执行,YYY这块资源就不会回收了,一直占用着,这样的代码一多,是可能引起资源句柄泄露的。而改为下面的写法之后,就保证了无论如何XXX和YYY都会被close掉。


Related labels:
source:php.cn
Statement of this Website
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn
Popular Tutorials
More>
Latest Downloads
More>
Web Effects
Website Source Code
Website Materials
Front End Template
About us Disclaimer Sitemap
php.cn:Public welfare online PHP training,Help PHP learners grow quickly!