Flash 优化ByteArray复制
下面是来自flashlight VNC库的代码片段,它解压缩(使用ByteArray.inflate)传入的字节流Flash 优化ByteArray复制,flash,actionscript-3,apache-flex,optimization,bytearray,Flash,Actionscript 3,Apache Flex,Optimization,Bytearray,下面是来自flashlight VNC库的代码片段,它解压缩(使用ByteArray.inflate)传入的字节流 package com.flashlight.zlib { import com.flashlight.utils.TimeTracker; import flash.utils.ByteArray; import mx.logging.ILogger; import mx.logging.Log; public class Infl
package com.flashlight.zlib
{
import com.flashlight.utils.TimeTracker;
import flash.utils.ByteArray;
import mx.logging.ILogger;
import mx.logging.Log;
public class Inflater {
private var lastDeflate:ByteArray;
public function uncompress(compressedData:ByteArray):ByteArray {
var uncompressedData:ByteArray = new ByteArray();
var dataOffset:int = lastDeflate ? 0 : 2;
TimeTracker.startTimer("TightEncoding[CopyCompresion]");
if (lastDeflate) {
var dictionarySize:int = lastDeflate.length > 32768 ? 32768 : lastDeflate.length;
uncompressedData.writeByte(0x00);
uncompressedData.writeByte(dictionarySize );
uncompressedData.writeByte(dictionarySize >> 8);
uncompressedData.writeByte(~dictionarySize);
uncompressedData.writeByte((~dictionarySize) >> 8 );
uncompressedData.writeBytes(lastDeflate,lastDeflate.length - dictionarySize, dictionarySize);
}
uncompressedData.writeBytes(compressedData,dataOffset,compressedData.length-dataOffset);
TimeTracker.stopTimer("TightEncoding[CopyCompresion]");
uncompressedData.writeByte(0x01);
uncompressedData.writeUnsignedInt(0x0000FFFF);
TimeTracker.startTimer("TightEncoding[Realdecompress]");
uncompressedData.inflate();
TimeTracker.stopTimer("TightEncoding[Realdecompress]");
lastDeflate = uncompressedData;
uncompressedData.position = dictionarySize;
return uncompressedData;
}
}
}
来自服务器的压缩数据流是恒定的,由充气器类逐块充气。在这个类中,膨胀()和writeBytes()方法总共占用99%的时间。inflate()已经是本机调用
如何优化writeBytes调用?我们可以跳过它,以其他方式重新连接代码,还是有更好的优化方法来执行ByteArray复制操作?不确定ByteArray是否可以优化,只要它是本机Flash类。我想你可以尝试一些第三方库/框架,比如氟或炼金术,调用C++算法,可以更快。我尝试过,对于CPU广泛的“to”循环,但是通过BurtRay-Rad需要大量的时间:(我现在正在寻找HAXE/AZOM,以提高BurtRayRayPipe性能。