【Eigen 下MKL使用】在Ubuntu系统下 基于Eigen使用MKL框架和openMP框架(Qt Creator)

参考链接：

1、https://software.intel.com/en-us/articles/intel-mkl-link-line-advisor/

2、https://blog.****.net/liang19890820/article/details/51774724

3、http://www.it1352.com/367668.html

4、http://eigen.tuxfamily.org/dox/TopicUsingIntelMKL.html

1、直接上代码，首先配置.pro文件。如下

QT += core
QT -= gui

CONFIG += c++11

TARGET = testEigenProcess
CONFIG += console
CONFIG -= app_bundle

TEMPLATE = app

SOURCES += main.cpp

# The following define makes your compiler emit warnings if you use
# any feature of Qt which as been marked deprecated (the exact warnings
# depend on your compiler). Please consult the documentation of the
# deprecated API in order to know how to port your code away from it.
DEFINES += QT_DEPRECATED_WARNINGS

# You can also make your code fail to compile if you use deprecated APIs.
# In order to do so, uncomment the following line.
# You can also select to disable deprecated APIs only up to a certain version of Qt.
#DEFINES += QT_DISABLE_DEPRECATED_BEFORE=0x060000    # disables all the APIs deprecated before Qt 6.0.0

# Eigen
INCLUDEPATH += \
/home/david/MySoft/myeigen
# openMP
QMAKE_CXXFLAGS += -fopenmp
LIBS += -fopenmp
# intel MKL
unix:INCLUDEPATH += /opt/intel/mkl/include
unix:LIBS += -L/opt/intel/mkl/lib/intel64 \
    -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core \
    -L/opt/intel/lib/intel64 \
    -liomp5 -lpthread -ldl -lm

2、测试代码如下：

#define EIGEN_USE_MKL_ALL

#include <QCoreApplication>
#include <Eigen/Dense>
#include <iostream>
#include <time.h>
#include <omp.h>


using namespace Eigen;
using namespace  std;

void testOpenMP();

void print_xiao(char i)
{
    cout << "omp_get_num_threads : " << omp_get_num_threads() << endl;
    cout << "omp_get_max_threads: " << omp_get_max_threads() << endl;
    cout << "omp_get_thread_num: " << omp_get_thread_num() << endl;
    cout << "omp_get_num_procs: " << omp_get_num_procs() << endl;
    cout << i << endl;
}

int main(int argc, char *argv[])
{
    QCoreApplication a(argc, argv);
    initParallel();
    // use openMP
    cout << "before set openMP threads: " << omp_get_num_threads() << endl;
    int set_thread = 10;
    omp_set_num_threads(set_thread);
    cout << "after set openMP threads: " << omp_get_num_threads() << endl;
//    #pragma omp parallel for
    for (char i = 'a'; i <= 'z'; i++)
        print_xiao(i);



    // test mult-threading
    int mat_size = 1000;
    MatrixXd matXd, save_mat;
    clock_t time_start = 0, time_end = 0;
    matXd.resize(mat_size, mat_size);
    matXd.setIdentity();
    // use parallel maybe Slower for each matrix, but the overall speed will be fast.
//    #pragma omp parallel for
    for(int i = 0; i < 10;i++)
    {
        time_start = clock();
//        save_mat = matXd.inverse();
        save_mat = matXd * matXd;
    //    cout << save_mat << endl;
        time_end = clock();
        cout <<i << " -> Elapsed time is "
                << (double)(time_end - time_start)/CLOCKS_PER_SEC<< " seconds." << endl;
    }




//    testOpenMP();
    return a.exec();
}


void testOpenMP()
{
    double x,y,i,j;
    long count=0,count1=0,count2=0;
    clock_t startTime = 0, endTime = 0;

    startTime=clock();

    #pragma omp parallel sections
    {
    #pragma omp section
        {
         for(x=0;x<=0.5;x+=0.0001)
            for(y=0;y<=1;y+=0.0001)
             {
             if(x*x+y*y<=1) count1++;
             }
        }
    #pragma omp section
       {
        for(i=0.5001;i<=1;i+=0.0001)
            for(j=0;j<=1;j+=0.0001)
             {
             if(i*i+j*j<=1) count2++;
             }
        }
    }
    count=count1+count2;
    endTime=clock();

    cout <<"testOpenMP() Elapsed time is "
            << (double)(endTime - startTime)/CLOCKS_PER_SEC<< " seconds." << endl;

    // no parallel
    startTime=clock();

    #pragma omp parallel sections
    {
    #pragma omp section
        {
         for(x=0;x<=0.5;x+=0.0001)
            for(y=0;y<=1;y+=0.0001)
             {
             if(x*x+y*y<=1) count1++;
             }
        }
    #pragma omp section
       {
        for(i=0.5001;i<=1;i+=0.0001)
            for(j=0;j<=1;j+=0.0001)
             {
             if(i*i+j*j<=1) count2++;
             }
        }
    }
    count=count1+count2;
    endTime=clock();

    cout <<"testOpenMP() Elapsed time is "
            << (double)(endTime - startTime)/CLOCKS_PER_SEC<< " seconds." << endl;
}

3、真多1000*1000矩阵想乘法，用不用#define EIGEN_USE_MKL_ALL 结果图如下：

不使用如下图：

使用MKL如下图：