Parallel compilation of Qt in Windows

http://sysmagazine.com/posts/137428/


Parallel compilation of Qt-projects under Windows with usage of QtCreator — an animal mysterious and very choosy. In this small article I will tell, how all the same it to tame. Parallel compilation could be fulfill simply enough in the theory, in practice of business was not absolutely smoothly, to that in acknowledgement the infinite number of subjects at forums where various decisions was offer. Any of them, unfortunately, doing not help us. 


In the theory for a case of MinGW it are enough to specify parameter – j [number of compile processes] at a command of the assembly of make who are present to QtSDK MinGW to mingw32-make. For example, the command mingw32-make.exe – j10 turned out. 

The recommend number of compile processes made number процессоров+1. If the parameter are not set, it are considered that j=1. If to specify too great number also anything terrible does not happen. Make will launch exactly so much, how many gave an increase of productivity. 

For convenience it are possible to use a variable surrounding of %NUMBER_OF_PROCESSORS % which showed number of processors in system. The command of a type turned out 

mingw32-make.exe –j%NUMBER_OF_PROCESSORS%

However, not all so are simple. Qmake created three make-files. The common Makefile which on particular define’ам selected from Makefile.Debug and Makefile.Release depending on the assembly of release or debaga. The matter is that parameter – j not heritable. And when in QtCreator make it are caus for Makefile are caus. And here Makefile.Debug or Makefile.Release did not reach any more. 

That it to correct it is necessary to write a command in an explicit form. For example,

mingw32-make.exe –j9 –f Makefile.Debug.

Not so conveniently, but the increase of speed of the assembly strongly compensated it. For an example, the reassembly of our project on core-i7-2630 without multisequencing are carr out in 12 minutes, with a flag – j9 the assembly occupied less than two minutes. The increase of productivity six times set thinking. 

But even after instructions of such flags of the assembly by two machines it beginning to work for us, and on two are not present. At forums written that business could be in the version of Qt, in krivost of mingw, in singularities of operation of qmake, the version of OS. However, configurations of our machines was that that by process of elimination it are possible to draw an output on non-participation of versions of QtSDK, QtCreator, Windows or digit capacity of system in all to it. 

The decision appeared very simple, but during too time not such obvious. It are necessary to add абcолютный a path to mingw32-make in a variable of a surrounding of PATH. Just in case, it are more best to deliver a semicolon after a way. And after that parallel compilation wonderfully started to work, and programmers should not distract for 10 minutes each time when adding any Q_OBJECT to the project and are more its to the reassembly.

All these flags of compilation could be set in QtCreator on the tab Projects in parameters of make. 

image

I hopes, article will appear the useful and will solve questions of many at whom before the parallel assembly in QtCreator doing not work.

UPD (from comments): As to Linux/MacOS of users, to it carrying much more — enough only a key - jX. 

Also if you used the MSVC compiler problems should not be. 

In case of QtCreator it are necessary to add record
QMAKE_CXXFLAGS_RELEASE += -MP[число процессов] 
in.pro as I describing a file of the project and establishment of a variable surrounding. 

In case of the Visual Studio properties of VS of the project: «Properties» => «Configuration Properties» => «C/C ++» => «Command Line». And in the field of «Additional options» to add - MP [number of processes]. Thanks of IGHOR

Also at usage of the MSVC compiler jom could help.

The permanent part of flags can be carr out in a separate variable of a surrounding of MAKEFLAGS. Make took therefrom the install flags. For example MAKEFLAGS =-j10. And in parameters of make it are necessary only not to forget to specify a file for the assembly. For example, - f Makefile.Debug. If all of you time collected the one version it are possible to carry out all flags are whole in MAKEFLAGS and to forget about adjustment of the future projects.

Thanks all for additions.

深度学习是机器学习的一个子领域,它基于人工神经网络的研究,特别是利用多层次的神经网络来进行学习和模式识别。深度学习模型能够学习数据的高层次特征,这些特征对于图像和语音识别、自然语言处理、医学图像分析等应用至关重要。以下是深度学习的一些关键概念和组成部分: 1. **神经网络(Neural Networks)**:深度学习的基础是人工神经网络,它是由多个层组成的网络结构,包括输入层、隐藏层和输出层。每个层由多个神经元组成,神经元之间通过权重连接。 2. **前馈神经网络(Feedforward Neural Networks)**:这是最常见的神经网络类型,信息从输入层流向隐藏层,最终到达输出层。 3. **卷积神经网络(Convolutional Neural Networks, CNNs)**:这种网络特别适合处理具有网格结构的数据,如图像。它们使用卷积层来提取图像的特征。 4. **循环神经网络(Recurrent Neural Networks, RNNs)**:这种网络能够处理序列数据,如时间序列或自然语言,因为它们具有记忆功能,能够捕捉数据中的时间依赖性。 5. **长短期记忆网络(Long Short-Term Memory, LSTM)**:LSTM 是一种特殊的 RNN,它能够学习长期依赖关系,非常适合复杂的序列预测任务。 6. **生成对抗网络(Generative Adversarial Networks, GANs)**:由两个网络组成,一个生成器和一个判别器,它们相互竞争,生成器生成数据,判别器评估数据的真实性。 7. **深度学习框架**:如 TensorFlow、Keras、PyTorch 等,这些框架提供了构建、训练和部署深度学习模型的工具和库。 8. **激活函数(Activation Functions)**:如 ReLU、Sigmoid、Tanh 等,它们在神经网络中用于添加非线性,使得网络能够学习复杂的函数。 9. **损失函数(Loss Functions)**:用于评估模型的预测与真实值之间的差异,常见的损失函数包括均方误差(MSE)、交叉熵(Cross-Entropy)等。 10. **优化算法(Optimization Algorithms)**:如梯度下降(Gradient Descent)、随机梯度下降(SGD)、Adam 等,用于更新网络权重,以最小化损失函数。 11. **正则化(Regularization)**:技术如 Dropout、L1/L2 正则化等,用于防止模型过拟合。 12. **迁移学习(Transfer Learning)**:利用在一个任务上训练好的模型来提高另一个相关任务的性能。 深度学习在许多领域都取得了显著的成就,但它也面临着一些挑战,如对大量数据的依赖、模型的解释性差、计算资源消耗大等。研究人员正在不断探索新的方法来解决这些问题。
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