Groovy 学习笔记3 运行效率
第一篇笔记里面,我说groovy运行的居然还满快的,其实是个误会了。我上次做八皇后还是在8080上面用basic做的,和现在奔四上面的groovy相比是没有意义的。特地又做了个对比试验:
1int q=9
2int[] i=new int[q]
3int count=0
4long t = System.currentTimeMillis();
5scan(0)
6println("totle results:"+count)
7println("totle time:"+(System.currentTimeMillis()-t));
8def scan(n){
9 if (n==q){
10 println(i.toList())
11 count++
12 return
13 }
14 i[n]=0
15 while(i[n]<q){
16 i[n] = i[n]+1
17 if (check(n))
18 scan(n+1)
19 }
20}
21def check(n){
22 if (n>0)
23 for (j in 0..<n)
24 if (i[j]==i[n] || i[j]-i[n]==j-n || i[j]-i[n]==n-j )
25 return false
26 return true
27}
2int[] i=new int[q]
3int count=0
4long t = System.currentTimeMillis();
5scan(0)
6println("totle results:"+count)
7println("totle time:"+(System.currentTimeMillis()-t));
8def scan(n){
9 if (n==q){
10 println(i.toList())
11 count++
12 return
13 }
14 i[n]=0
15 while(i[n]<q){
16 i[n] = i[n]+1
17 if (check(n))
18 scan(n+1)
19 }
20}
21def check(n){
22 if (n>0)
23 for (j in 0..<n)
24 if (i[j]==i[n] || i[j]-i[n]==j-n || i[j]-i[n]==n-j )
25 return false
26 return true
27}
运行结果是:totle time:7271 (为了用groovy控制台运行的,直接用groovy命令运行还要慢一点)
java呢?
queens.java:
1public class queens {
2 static int q=9;
3 static int[] i=new int[q];
4 static int count=0;
5 public static void main(String[] args){
6 long t = System.currentTimeMillis();
7 scan(0);
8 System.out.println("totle results:"+count);
9 System.out.println("totle time:"+(System.currentTimeMillis()-t));
10 }
11 private static void scan(int n){
12 if (n==q){
13 for (int k=0;k<q;k++) System.out.print(i[k]+(k==q-1?"/n":","));
14 count++;
15 return;
16 }
17 i[n]=0;
18 while(i[n]<q){
19 i[n] = i[n]+1;
20 if (check(n)){
21 scan(n+1);
22 }
23 }
24 }
25 private static boolean check(int n){
26 for(int j=0;j<n;j++){
27 if (i[j]==i[n] || i[j]-i[n]==j-n || i[j]-i[n]==n-j ){
28 return false;
29 }
30 }
31 return true;
32 }
33}
34
2 static int q=9;
3 static int[] i=new int[q];
4 static int count=0;
5 public static void main(String[] args){
6 long t = System.currentTimeMillis();
7 scan(0);
8 System.out.println("totle results:"+count);
9 System.out.println("totle time:"+(System.currentTimeMillis()-t));
10 }
11 private static void scan(int n){
12 if (n==q){
13 for (int k=0;k<q;k++) System.out.print(i[k]+(k==q-1?"/n":","));
14 count++;
15 return;
16 }
17 i[n]=0;
18 while(i[n]<q){
19 i[n] = i[n]+1;
20 if (check(n)){
21 scan(n+1);
22 }
23 }
24 }
25 private static boolean check(int n){
26 for(int j=0;j<n;j++){
27 if (i[j]==i[n] || i[j]-i[n]==j-n || i[j]-i[n]==n-j ){
28 return false;
29 }
30 }
31 return true;
32 }
33}
34
运行结果是:totle time:271
每次运行花费的时间略有不同,groovy和java的运行速度看来大致相差10~30倍左右。
能说这是脚本语言天生的缺陷吗?我们来看看同样是类似java语法的脚本语言javascript在IE里面的速度:
1var q=9
2var i=[]
3var count=0
4var d = new Date();
5scan(0)
6document.write("totle results:"+count+"<br>")
7document.write("time used:"+(new Date()-d)+"<br>")
8
9function scan(n){
10 if (n==q){
11 document.write(i+"<br>")
12 count++
13 return
14 }
15 i[n]=0
16 while(i[n]<q){
17 i[n] = i[n]+1
18 if (check(n)){
19 scan(n+1)
20 }
21 }
22}
23
24function check(n){
25 for (var j=0; j<n;j++)
26 if (i[j]==i[n] || i[j]-i[n]==j-n || i[j]-i[n]==n-j )
27 return false
28 return true
29}
2var i=[]
3var count=0
4var d = new Date();
5scan(0)
6document.write("totle results:"+count+"<br>")
7document.write("time used:"+(new Date()-d)+"<br>")
8
9function scan(n){
10 if (n==q){
11 document.write(i+"<br>")
12 count++
13 return
14 }
15 i[n]=0
16 while(i[n]<q){
17 i[n] = i[n]+1
18 if (check(n)){
19 scan(n+1)
20 }
21 }
22}
23
24function check(n){
25 for (var j=0; j<n;j++)
26 if (i[j]==i[n] || i[j]-i[n]==j-n || i[j]-i[n]==n-j )
27 return false
28 return true
29}
运行结果是: time used:1241
比groovy快了5倍以上。groovy可真是够慢的。
把groovy编译的class文件反编译了一下,看到groovy生成的代码效率确实是太低了,我们就看循环最内层的check函数吧:
1def check(n){
2 if (n>0)
3 for (j in 0..<n)
4 if (i[j]==i[n] || i[j]-i[n]==j-n || i[j]-i[n]==n-j )
5 return false
6 return true
7}
2 if (n>0)
3 for (j in 0..<n)
4 if (i[j]==i[n] || i[j]-i[n]==j-n || i[j]-i[n]==n-j )
5 return false
6 return true
7}
编译后变成
1 public Object check(Object obj)
2 {
3 if(ScriptBytecodeAdapter.compareGreaterThan(obj, new Integer(0)))
4 {
5 Object obj1 = null;
6 for(Iterator iterator = ScriptBytecodeAdapter.asIterator(ScriptBytecodeAdapter.createRange(new Integer(0), obj, false)); iterator.hasNext();)
7 {
8 Object obj2 = iterator.next();
9 Object obj3 = null;
10 if(ScriptBytecodeAdapter.asBool(ScriptBytecodeAdapter.asBool(ScriptBytecodeAdapter.compareEqual(ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
11 obj2
12}))), ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
13 obj
14})))) || ScriptBytecodeAdapter.compareEqual(ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
15 obj2
16}))), "minus", ((Object) (new Object[] {
17 ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
18 obj
19 })))
20}))), ScriptBytecodeAdapter.invokeMethod(obj2, "minus", ((Object) (new Object[] {
21 obj
22})))) ? ((Object) (Boolean.TRUE)) : ((Object) (Boolean.FALSE))) || ScriptBytecodeAdapter.compareEqual(ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
23 obj2
24}))), "minus", ((Object) (new Object[] {
25 ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
26 obj
27 })))
28}))), ScriptBytecodeAdapter.invokeMethod(obj, "minus", ((Object) (new Object[] {
29 obj2
30})))) ? ((Object) (Boolean.TRUE)) : ((Object) (Boolean.FALSE))))
31 return Boolean.FALSE;
32 }
33
34 }
35 return Boolean.TRUE;
36 }
37
2 {
3 if(ScriptBytecodeAdapter.compareGreaterThan(obj, new Integer(0)))
4 {
5 Object obj1 = null;
6 for(Iterator iterator = ScriptBytecodeAdapter.asIterator(ScriptBytecodeAdapter.createRange(new Integer(0), obj, false)); iterator.hasNext();)
7 {
8 Object obj2 = iterator.next();
9 Object obj3 = null;
10 if(ScriptBytecodeAdapter.asBool(ScriptBytecodeAdapter.asBool(ScriptBytecodeAdapter.compareEqual(ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
11 obj2
12}))), ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
13 obj
14})))) || ScriptBytecodeAdapter.compareEqual(ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
15 obj2
16}))), "minus", ((Object) (new Object[] {
17 ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
18 obj
19 })))
20}))), ScriptBytecodeAdapter.invokeMethod(obj2, "minus", ((Object) (new Object[] {
21 obj
22})))) ? ((Object) (Boolean.TRUE)) : ((Object) (Boolean.FALSE))) || ScriptBytecodeAdapter.compareEqual(ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
23 obj2
24}))), "minus", ((Object) (new Object[] {
25 ScriptBytecodeAdapter.invokeMethod(ScriptBytecodeAdapter.getGroovyObjectProperty(this, "i"), "getAt", ((Object) (new Object[] {
26 obj
27 })))
28}))), ScriptBytecodeAdapter.invokeMethod(obj, "minus", ((Object) (new Object[] {
29 obj2
30})))) ? ((Object) (Boolean.TRUE)) : ((Object) (Boolean.FALSE))))
31 return Boolean.FALSE;
32 }
33
34 }
35 return Boolean.TRUE;
36 }
37
一切都是object,做任何事情都是invokeMethod,两个整数的比较居然要写将近400个字符的代码,光看代码量都可以吓倒我了。这是我们期待的脚本语言吗?
groovy可以嵌入到java代码里面,但是java代码可以嵌入到groovy里面吗?我觉得groovy有必要提供这样一种机制,在有必要的时候可以消除性能瓶颈。可是现在只看到groovy里面可以通过Scriptom(现在还是beta版)嵌入vbs、js脚本(包括使用WSH,FSO)或者调用InternetExplorer、Media Player、Word和Excel等windows组件。看来对消除性能瓶颈的帮助不大。