KMThreadPool: 6 – Thread Pool Step 4:How to Use

6 – Thread Pool Step 4:How to Use

This is just a quick overview onwhat you need to do to use the thread pool.

Note: In this tutorial, I’ve shown you functionsthat you need to create the thread pool. There are, however, smaller functions(accessors, mutators, etc.) that I haven’t shown you, but are easy tounderstand.

Note 2: The example project also was compiledusing VisualLeak Detector. If you try to run the sample without VLD, you’ll get acrash. Just comment out the #include in “main.cpp” or go and download the handy-dandytool.

Adding the Thread Pool To YourProject

This is also expressed in theREADME that comes with the sample project in on the download page for thisproject. Really it’s quite simple, and there are 2 ways to go about it.

Variant 1: Add the library to your project.

This method is a bit hard to getdown, but when you do it enough times, you’ll get it. And it’s always righthere to reference back to.

  1. Cut a hole in a box
  2. Put your j–

WHOA! Wrong instructions!

Damn you,Timberlake!

OK, here they are…

  1. Put the libKMThreadPool folder from the sample project somewhere on your computer that’s easy to reference to (for example: “C:/C++/libKMThreadPool/”
  2. Go to Visual Studios and open the solution that contains your project (lets call it “MyProj”)
  3. Include the project file for libKMThreadPool into your solution
  4. Right-click on MyProj and open Properties
  5. In “C++>>General” under “Additional Include Directories,” add in the path to the thread pools’ directory (in this example it would be “C:/C++/libKMThreadPool/”)
  6. Now, still in Properties, go in “Linker>>Input” under “Additional Dependencies” and add libKMThreadPool.lib
  7. Now exit Properties, click on MyProj, and click on Project Dependencies.
  8. Check the checkbox that reads “libKMThreadPool”

That’s all there is to it! If youcompile the project now, the output .lib file from libKMThreadPool will appearin the same directory as your solution for MyProj is in. Cool, huh?

Variant 2: Add the sourcedirectly into your project.

This is real easy. If for somereason the above method doesn’t work, then just copy all the source files fromlibKMThreadPool and chuck them directly into your project. Quick and dirty.

Initializing and Shutting Downthe Thread Pool

Including the files

The thread pool I have set up areall in namespaces. The primary namespace is kmp. The threading stuff (such as KMLock,KMThread, etc.) are all stored in kmp::threading, any Utilities are under kmp::threading::utility, and any algorithms are in kmp::threading::algorithms.

Since you included thedirectories, you won’t need some long, horrible-looking #include. The onlyfiles you’ll really be interacting with a lot are KMThreadPool_Win32.h” and“KMUtility.h” (and “KMLock.h” maybe).

Initializing

First thing’s first – when youwant to use the thread pool from anywhere, you’ve got to grab the singletoninstance of it.

KMThreadPool* pthreadpool = KMThreadPool::getInstance();

That’s easy. If you’re doing thisfrom inside a class definition, you may want to store the instance so you’renot constantly calling getInstance(). Once you have that, now it’s time to callinitialize().

But wait! I know you’re ready tocall pthreadpool->Initialize(1000, 1000000); but that would be foolish! You processordoesn’t have that much power, but it will try to create at least 1000 threads,but maybe only 4 of them will be actual threads! All the rest will be virtualthreads.

Most processors are designedonly to really allow 2 threads per core (some have been known to go as farup as 8), so reasonably, if you have a dual-core processor, your max threadsshould be 4, and min should be 2. If you try to increase that number, you’llnotice your program will run very sluggish.

Now that our thread pool isready, we can call BeginProcessing() and start adding tasks!

Shut down

Once your program is finished,it’s a good idea to clean up the memory this thread pool has allocated. Now,you can call StopProcessing() and then call Shutdown() on the thread pool, or you can just call Shutdown(). Either way, call Shutdown() once your program is finished. Done.

Adding Tasks

 Adding tasks require a bitof set up. The first thing we want to do is set up a child class of IKMTaskData. This class is going to contain dataneeded for your task, like, say, float fElapsedTime, int Index, GameObject* curObj, etc, etc. See the examples in the sampleproject if you need more explanation, but there really isn’t much to explain here.

Now for the task function itself.Remember: the task function must be static, as explained in Step 2. This taskmust be set up just like KMTaskFunc – same return value, same parameter list. If youwant to access your data you made in the child of IKMTaskData from the IKMTaskData*, you have to cast the IKMTaskData* to your child class, also as explained inStep 2.

So lets say we have voidMyTask(IKMTaskData*)and MyTaskData which derives from IKMTaskData and contains 2 ints and a constructor. How do I getthis to the pool?

pthreadpool->AddTask(MyTask, new MyTaskData(10, 67));

Easy.

转自:http://keithmaggio.wordpress.com/code/c-win32-thread-pool-manager/6-step4/

  • 0
    点赞
  • 0
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
目标检测(Object Detection)是计算机视觉领域的一个核心问题,其主要任务是找出图像中所有感兴趣的目标(物体),并确定它们的类别和位置。以下是对目标检测的详细阐述: 一、基本概念 目标检测的任务是解决“在哪里?是什么?”的问题,即定位出图像中目标的位置并识别出目标的类别。由于各类物体具有不同的外观、形状和姿态,加上成像时光照、遮挡等因素的干扰,目标检测一直是计算机视觉领域最具挑战性的任务之一。 二、核心问题 目标检测涉及以下几个核心问题: 分类问题:判断图像中的目标属于哪个类别。 定位问题:确定目标在图像中的具体位置。 大小问题:目标可能具有不同的大小。 形状问题:目标可能具有不同的形状。 三、算法分类 基于深度学习的目标检测算法主要分为两大类: Two-stage算法:先进行区域生成(Region Proposal),生成有可能包含待检物体的预选框(Region Proposal),再通过卷积神经网络进行样本分类。常见的Two-stage算法包括R-CNN、Fast R-CNN、Faster R-CNN等。 One-stage算法:不用生成区域提议,直接在网络中提取特征来预测物体分类和位置。常见的One-stage算法包括YOLO系列(YOLOv1、YOLOv2、YOLOv3、YOLOv4、YOLOv5等)、SSD和RetinaNet等。 四、算法原理 以YOLO系列为例,YOLO将目标检测视为回归问题,将输入图像一次性划分为多个区域,直接在输出层预测边界框和类别概率。YOLO采用卷积网络来提取特征,使用全连接层来得到预测值。其网络结构通常包含多个卷积层和全连接层,通过卷积层提取图像特征,通过全连接层输出预测结果。 五、应用领域 目标检测技术已经广泛应用于各个领域,为人们的生活带来了极大的便利。以下是一些主要的应用领域: 安全监控:在商场、银行
目标检测(Object Detection)是计算机视觉领域的一个核心问题,其主要任务是找出图像中所有感兴趣的目标(物体),并确定它们的类别和位置。以下是对目标检测的详细阐述: 一、基本概念 目标检测的任务是解决“在哪里?是什么?”的问题,即定位出图像中目标的位置并识别出目标的类别。由于各类物体具有不同的外观、形状和姿态,加上成像时光照、遮挡等因素的干扰,目标检测一直是计算机视觉领域最具挑战性的任务之一。 二、核心问题 目标检测涉及以下几个核心问题: 分类问题:判断图像中的目标属于哪个类别。 定位问题:确定目标在图像中的具体位置。 大小问题:目标可能具有不同的大小。 形状问题:目标可能具有不同的形状。 三、算法分类 基于深度学习的目标检测算法主要分为两大类: Two-stage算法:先进行区域生成(Region Proposal),生成有可能包含待检物体的预选框(Region Proposal),再通过卷积神经网络进行样本分类。常见的Two-stage算法包括R-CNN、Fast R-CNN、Faster R-CNN等。 One-stage算法:不用生成区域提议,直接在网络中提取特征来预测物体分类和位置。常见的One-stage算法包括YOLO系列(YOLOv1、YOLOv2、YOLOv3、YOLOv4、YOLOv5等)、SSD和RetinaNet等。 四、算法原理 以YOLO系列为例,YOLO将目标检测视为回归问题,将输入图像一次性划分为多个区域,直接在输出层预测边界框和类别概率。YOLO采用卷积网络来提取特征,使用全连接层来得到预测值。其网络结构通常包含多个卷积层和全连接层,通过卷积层提取图像特征,通过全连接层输出预测结果。 五、应用领域 目标检测技术已经广泛应用于各个领域,为人们的生活带来了极大的便利。以下是一些主要的应用领域: 安全监控:在商场、银行
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值