多线程效率问题

Question

我在项目中使用多线程处理数据遇到一些问题，请大家帮忙看看
具体实现如下：
一个主控制函数Main(),它主要用来分配线程参数和控制线程的启动
Main()示例代码如下：
for (int i = 0; i < threadNum; i++)
{
    ......根据线程数分配参数：beginTime,endTime(代码省略）    string[] objGpsTime= {beginGPSTime.ToString("HHmmss"), endGPSTime.ToString("HHmmss")};
   //启动线程
   ThreadPool.QueueUserWorkItem(new WaitCallback(AnalyseData), objGpsTime);
}
一个数据处理函数AnalyseData(beginTime,endTime),它主要用来处理不同时间段的数据
AnalyseData主要做以下动作：
1、根据传入的时间参数从数据库中取相关数据
2、遍历取出的数据进行处理（主要是做一些空间分析）我发现不管用多少个线程最终数据处理完成的时间都是一样的
请大家帮忙分析一下什么原因

Answer 1

“AnalyseData主要做以下动作： 
1、根据传入的时间参数从数据库中取相关数据 ”数据库是否加了锁，什么样的锁，是否会导致多个线程实际上的线性执行？
PS：最好是将数据一次头读出来，然后多线程来处理这些数据。

Answer 2

for (int i = 0; i < threadNum; i++) 
{ 
    ......根据线程数分配参数：beginTime,endTime(代码省略）     string[] objGpsTime= {beginGPSTime.ToString("HHmmss"), endGPSTime.ToString("HHmmss")}; 
  //启动线程 
  ThreadPool.QueueUserWorkItem(new WaitCallback(AnalyseData), objGpsTime); 
  //时间片要注意释放
  Thread.Sleep(0);
} 

Answer 3

锁在AnalyseData方法中互斥，对一些共享的数据进行保护就可以了。保护的代码越少效率越高。

Answer 4

我把整个代码贴出来大家帮忙看看~
public void Main(int threadNum)
        {
            Thread[] threads = new Thread[threadNum];            OnlineFcdDs fcdDs = new OnlineFcdDs();
            FCDCarStatusDs carStatusDs = new FCDCarStatusDs();
            PubRoadConfigDs roadCfgDs = new PubRoadConfigDs();
            //取当前数据库服务器时间
            //DateTime currentDbTime = DateTime.Now;
            DateTime currentDbTime = new DateTime(2009, 10, 8, 00, 08, 00);
            //数据过滤
            //fcdBll.OnlineFilter(beginGPSTime, endGPSTime);            //如果线程数大于当前时间间隔则将线程数置为当前时间间隔的1/4
            if (threadNum >= AnalyseEdgeStatus_Interval)
            {
                threadNum = (int)(AnalyseEdgeStatus_Interval / 4);
            }            // 获取线程池的最大线程数和维护的最小空闲线程数
            int maxThreadNum, portThreadNum;
            int minThreadNum;
            ThreadPool.GetMaxThreads(out maxThreadNum, out portThreadNum);
            ThreadPool.GetMinThreads(out minThreadNum, out portThreadNum);
            WriteLog(string.Format("最大线程数：", maxThreadNum));
            WriteLog(string.Format("最小空闲线程数：", minThreadNum));            //根据线程数分割时间节点
            int perGPSTime = AnalyseEdgeStatus_Interval / threadNum;
            int remNum = AnalyseEdgeStatus_Interval % threadNum;
            DateTime beginGPSTime, endGPSTime;
            for (int i = 0; i < threadNum; i++)
            {
                if (i == 0)
                {
                    beginGPSTime = currentDbTime.AddSeconds(i * perGPSTime);
                }
                else
                {
                    beginGPSTime = currentDbTime.AddSeconds(i * perGPSTime + i);
                }
                //if (i==threadNum-1 && remNum > 0)
                //{
                //    endGPSTime = beginGPSTime.AddSeconds(perGPSTime + remNum);
                //}
                if (i == threadNum - 1)
                {
                    endGPSTime = currentDbTime.AddSeconds(AnalyseEdgeStatus_Interval);
                }
                else
                {
                    endGPSTime = beginGPSTime.AddSeconds(perGPSTime);
                }
                //调用地址匹配线程
                //Thread fcdThread = new Thread(delegate() { AnalyseFCDThread(beginGPSTime.ToString("HHmmss"), endGPSTime.ToString("HHmmss")); });
                string[] objGpsTime= {beginGPSTime.ToString("HHmmss"), endGPSTime.ToString("HHmmss")};
                //Thread fcdThread = new Thread(new ParameterizedThreadStart(AnalyseFCDThread));
                //fcdThread.Name = string.Format("Thread_{0}", i.ToString());
                //fcdThread.Start(objGpsTime);                //启动线程
                ThreadPool.QueueUserWorkItem(new WaitCallback(AnalyseFCDThread), objGpsTime);
            }
            Console.ReadLine();
        }

Answer 5

private void AnalyseFCDThread(object objGpsTime)
        {
            WriteLog(string.Format("线程：{0} 开始启动......", Thread.CurrentThread.GetHashCode()));
            string[] gpsTime = objGpsTime as string[];
            string beginGPSTime = gpsTime[0];
            string endGPSTime = gpsTime[1];
            OnlineFcdDs fcdDs = new OnlineFcdDs();
            FCDCarStatusDs carStatusDs = new FCDCarStatusDs();
            perTimer.Start();
            //取过滤完成的FCD数据
            fcdBll.GetOnlineFcdData(fcdDs, beginGPSTime, endGPSTime);            //进行地址匹配（经纬度匹配）。
            //1、遍历FCD点进行最短距离计算
            //2、进行方位角判断
            double MinDistance = 0;
            double DistanceAlongCurve = 0;
            IPoint fromPoint = new PointClass();
            Boolean bAsRatio = false;
            Boolean bRight = false;
            //与缓冲区有交集的并且方位角夹角小于特定值的线集合
            Dictionary<IFeature, double> roadLine = new Dictionary<IFeature, double>();            foreach (DataRow dr in fcdDs.T_ONLINEFCD.Rows)
            {
                double tx, ty;
                if (double.TryParse(dr["LONGITUDE"].ToString(), out tx) && double.TryParse(dr["LATITUDE"].ToString(), out ty))
                {
                    IPoint fcdPoint = new PointClass();
                    fcdPoint.SpatialReference = ispRef;
                    fcdPoint.PutCoords(tx, ty);
                    //构造缓冲区
                    IGeometry bufferGeo = gisHp.GetBufferGeometry(fcdPoint as IGeometry, bufferDistance);
                    //求出与该缓冲区有交集的所有路段
                    List<IFeature> listLines = gisHp.GetIntersectGeometry(bufferGeo, pLines);
                    //判断方位角
                    double fcdAngle;
                    double.TryParse(dr["Rate"].ToString(), out fcdAngle);
                    if (listLines != null)
                    {
                        roadLine.Clear();
                        foreach (IFeature pFeature in listLines)
                        {
                            double lineAngle;
                            double.TryParse(pFeature.get_Value(pFeature.Fields.FindField("ANGLE")).ToString(), out lineAngle);
                            double difAngle = Math.Abs(lineAngle - fcdAngle);
                            //double difAngle = Math.Abs(tangLineAngle - fcdAngle);
                            if (difAngle >= 0) // <= 65)
                            {
                                roadLine.Add(pFeature, MinDistance);
                            }
                        }
                        if (roadLine.Count > 0)
                        {
                            //根据最短距离对线集合进行排序(升序)
                            var sortLines = from disValue in roadLine orderby disValue.Value select disValue;
                            IFeature edgeLine = null;
                            foreach (KeyValuePair<IFeature, double> pair in sortLines)
                            {
                                int edgeID;
                                double edgeLength;
                                decimal speed;
                                edgeLine = pair.Key as IFeature;
                                Int32.TryParse(edgeLine.get_Value(edgeLine.Fields.FindField("EDGEID")).ToString(), out edgeID);
                                double.TryParse(edgeLine.get_Value(edgeLine.Fields.FindField("LENGTH")).ToString(), out edgeLength);
                                if (edgeID > 0 && edgeLength > 0)
                                {
T_OnLineFCD_CarStatus
                                    DataRow carStatusDr = carStatusDs.T_ONLINEFCD_CARSTATUS.NewRow();
                                    carStatusDr["TERMINAL"] = dr["TERMINAL"];
                                    decimal.TryParse(dr["SPEED"].ToString(), out speed);
                                    carStatusDr["SPEED"] = decimal.Round(speed, 3);
                                    carStatusDr["EDGE_ID"] = edgeID;
                                    carStatusDr["EDGE_LENGTH"] = edgeLength;
                                    carStatusDr["GPSTIME"] = dr["GPSTIME"];
                                    carStatusDr["UPDATE_DATE"] = DateTime.Now.ToString("yyyyMMdd");
                                    carStatusDr["UPDATE_TIME"] = DateTime.Now.ToString("HHmmss");
                                    carStatusDr["FLAG"] = dr["FLAG"];
                                    //carStatusDr["SATELLITER_COUNT"] = dr[""];
                                    //其他列赋值
                                    carStatusDs.T_ONLINEFCD_CARSTATUS.Rows.Add(carStatusDr);
                                }
                                break;
                            }
                        }
                    }
                }
                dr.Delete();
            }
            perTimer.Stop();
            WriteLog(string.Format("{5}线程结束，时间间隔：{0}-{1}，FCD总数：{2}，匹配成功：{3}条，耗时：{4}",
                beginGPSTime, endGPSTime,
                fcdDs.T_ONLINEFCD.Rows.Count.ToString(), carStatusDs.T_ONLINEFCD_CARSTATUS.Rows.Count.ToString(),
                perTimer.Duration, Thread.CurrentThread.GetHashCode()));
            //停止当前线程
            Thread.CurrentThread.Abort();
            Console.ReadLine();
        }

Answer 6

perTimer 没在各线程里定义，在哪定义的？