ParallelProgrammingBasics_v2.pdf

Parallel Programming Basics
Jimmy Hu

Target Audience
 People interests parallel programming topic
 People wants to know how to improve the performance of their code
 People wants to know how to acquire the (possible) peak performance from their computer
(There are a bunch of techniques / methods available for reaching peak performance and this
kind of things is out of the range of our discussion)
 Someone wants to know the way that I am using my computers / servers (X
2

Outline
 Why parallel programming?
 What is parallel programming?
 How to perform parallel programming (in C++ / Matlab / C#)
 Conclusion / Further Discussions
3

Why Parallel Programming?
4
 Please check the following C++ code, what’s the output?
int main()
{
std::vector<int> test_vector = {1, 2, 3};
int a = 3;
for(int i = 0; i < std::ranges::size(test_vector); ++i)
{
test_vector[i] = test_vector[i] + a;
}
{
std::cout << test_vector[i] << ", ";
}
return 0;
}

5
 Answer of the question “Please check the following C++ code, what’s the output?”
// https://godbolt.org/z/fb7TdT495
// https://godbolt.org/z/3Kj1azb4h
int main()
{
int a = 3;
{
}
{
}
return 0;
}
4, 5, 6,

6
 Code Structure
int main()
{
int a = 3;
{
}
{
}
return 0;
}
Part 1: Variable Initialization

7
 Code Structure
int main()
{
int a = 3;
{
}
{
}
return 0;
}
Part 2: Data Processing / Calculation

8
 Code Structure
int main()
{
int a = 3;
{
}
{
}
return 0;
}
Part 3: Output

9
 In the mentioned simple example, the calculating part is simple add operation
int main()
{
int a = 3;
{
}
{
}
return 0;
}
4, 5, 6,

10
 What happened in the case of more complicated operation?
int main()
{
std::vector<Frame> image_frames =
{img1, img2, img3};
auto results = std::vector<Features>(3);
for(int i = 0; i < std::ranges::size(image_frames); ++i)
{
results[i] = feature_extraction(image_frames[i]);
}
…
return 0;
}
This example is calling a function
which named “feature_extraction”.

11
Without Parallel Programming With Parallel Programming
Dish A
Dish B
Dish C
…
Dish A Dish B Dish C
Icon is from https://www.hiclipart.com/free-transparent-background-png-clipart-iuxpq/download

12
Without Parallel Programming With Parallel Programming
Task A
Task B
Task C
…
Task A Task B Task C
Icon is from https://www.flaticon.com/free-icon/cpu_1250593

The Steps of Execution
13
int main()
{
int a = 3;
{
}
{
}
return 0;
}
 Let’s review the previous simple case. How’s the program is executed?

14
int main()
{
int a = 3;
{
}
{
}
return 0;
}
1 2 3
test_vector

15
1 2 3
test_vector
3
a
int main()
{
int a = 3;
{
}
{
}
return 0;
}

16
1 2 3
test_vector
3
a
 Then, the execution runs sequentially?
int main()
{
int a = 3;
{
}
{
}
return 0;
}

17
1 2 3
test_vector
3
a
= 4
int main()
{
int a = 3;
{
}
{
}
return 0;
}

18
4 2 3
test_vector
3
a
= 5
int main()
{
int a = 3;
{
}
{
}
return 0;
}

19
4 5 3
test_vector
3
a
= 6
int main()
{
int a = 3;
{
}
{
}
return 0;
}

20
4 5 6
test_vector
3
a
int main()
{
int a = 3;
{
}
{
}
return 0;
}

The Concept of Parallelization
21
1 2 3
test_vector
3
 Despite the way of sequentialization, is it possible to speed up?
 Why not let’s make the program runs parallelly (enable the operations run simultaneously)?
a 3
3
New
test_vector 4 5 6

22
1 2 3
test_vector
3
 How this can be done in our program? Solution: Parallel Programming!
a 3
3
New
test_vector 4 5 6

23
 Parallelization enabling
 Tools in C++:
- OpenMP
- TBB(Threading Building Blocks)
- std::thread
- Execution Policy in STL
 Tools in Matlab
 Tools in C#

Parallelization Implementation
24
 Parallelization enabling with OpenMP #include <omp.h>
int main()
{
auto test_vector = std::vector<int>{1, 2, 3};
int a = 3;
#pragma omp parallel for
for(int i = 0; i < test_vector.size(); i++)
{
}
for(int i = 0; i < test_vector.size(); ++i)
{
}
return 0;
}
int a = 3;
test_vector[0] =
test_vector[0] + a;
test_vector[1] =
test_vector[1] + a;
test_vector[2] =
test_vector[2] + a;
{
}
return 0;

25
 Parallelization enabling with OpenMP / TBB
// https://godbolt.org/z/szMc4jbqn
// https://godbolt.org/z/haM1qd6eY
#include <omp.h>
int main()
{
int a = 3;
{
}
{
}
return 0;
}
// https://godbolt.org/z/dcssoWj8K
#include <tbb/parallel_for.h>
int main()
{
int a = 3;
tbb::parallel_for( tbb::blocked_range<int>(0,test_vector.size()),
[&](tbb::blocked_range<int> r)
{
for (int i=r.begin(); i<r.end(); ++i)
{
}
});
{
}
return 0;
}

26
 Parallelization enabling with OpenMP / TBB
// https://godbolt.org/z/haM1qd6eY
#include <omp.h>
int main()
{
int a = 3;
for(int i = 0; i < test_vector.size(); i++)
{
}
{
}
return 0;
}
// https://godbolt.org/z/dcssoWj8K
#include <tbb/parallel_for.h>
int main()
{
int a = 3;
tbb::parallel_for( tbb::blocked_range<int>(0,test_vector.size()),
[&](tbb::blocked_range<int> r)
{
for (int i=r.begin(); i<r.end(); ++i)
{
}
});
{
}
return 0;
}
A lambda function is here!

Parallelization Methods Comparison
27
 Comparing OpenMP / TBB and std::thread
// The following code is an example of std::thread
// https://stackoverflow.com/a/11229853/6667035
// https://godbolt.org/z/YeY9d4EeP
#include <string>
#include <iostream>
#include <numeric>
#include <thread>
#include <vector>
void task1(std::vector<int> input) // The function we want to execute on the new thread.
{
for(int i = 0; i < input.size(); i++)
{
std::cout << "output from task1 function: " << input[i];
}
}
void function1()
{
auto test_vector1 = std::vector<int>(100);
std::iota(test_vector1.begin(), test_vector1.end(), 1);
int sum = 0;
for(int i = 0; i < test_vector1.size(); i++)
{
sum += test_vector1[i];
}
std::cout << sum << "n”;
}
int main()
{
auto test_vector = std::vector<int>(100);
std::iota(test_vector.begin(), test_vector.end(), 1);
std::thread t1(task1, test_vector);
function1();
t1.join();
return 0;
}

std::thread Concept
28
 Comparing OpenMP / TBB and std::thread
main function
task1 function
function1 function
Parallel Part
end

Execution Policy in STL
29
 When it comes to Execution Policy after C++17…
 std::execution::par
 std::execution::seq
// https://en.cppreference.com/w/cpp/algorithm/transform
// https://godbolt.org/z/bY14q1z3K
#include <algorithm>
#include <execution>
#include <iomanip>
#include <iostream>
#include <string>
#include <thread>
int main()
{
std::string g {"hello"};
std::for_each(std::execution::par, g.begin(), g.end(), [](char& c) // modify in-place
{
c = std::toupper(static_cast<unsigned char>(c));
});
std::cout << "g = " << std::quoted(g) << 'n';
return 0;
}

Parallelization in Matlab
30
 Document of parfor function usage

Parallelization in Matlab
31
 parfor function usage example
Program without parfor Program with parfor
// https://www.mathworks.com/help/parallel-
computing/parfor.html
tic
n = 200;
A = 500;
a = zeros(1,n);
for i = 1:n
a(i) = max(abs(eig(rand(A))));
end
toc
Elapsed time is 31.935373 seconds.
tic
n = 200;
A = 500;
a = zeros(1,n);
parfor i = 1:n
a(i) = max(abs(eig(rand(A))));
end
toc
Elapsed time is 10.760068 seconds.

Parallelization in C#
32
 Document of Parallel.For function usage: https://learn.microsoft.com/en-
us/dotnet/standard/parallel-programming/data-parallelism-task-parallel-library

Parallelization in C#
33
 Parallel.For function usage example
Program without Parallel.For Program with Parallel.For
using System;
using System.Threading.Tasks;
public class ParallelTest
{
public static void Main(string[] args)
{
for(int i = 0; i < 10; i++)
{
Console.WriteLine (i + "n");
};
}
}
using System;
using System.Threading.Tasks;
public class ParallelTest
{
public static void Main(string[] args)
{
Parallel.For(0, 10, i =>
{
Console.WriteLine (i + "n");
}); // Parallel.For
}
}
A lambda function is here!

Concept of Parallelable
35
 Please think that what’s the limitation of Parallelization

36
Answer: The limitation of parallelization is that the operation which is to be parallelize
should be independent!
What’s the meaning of independent?

37
Let’s check the case of dependent first:
A B C

38
Let’s check the case of dependent first:
The A, B and C operations cannot be made in parallelization!
Because B operation needs the output from A and
C operation needs the output from B!
A B C

Conclusion / Further Discussions
39
 Parallelization technique can bring some performance increment when you use it
properly
 Parallelization can make higher utilization of computers / computing devices
 Is there any disadvantage of using parallelization method?

Conclusion / Further Discussions
40
 Parallelization technique can bring some performance increment when you use it
properly
 Parallelization can make higher utilization of computers / computing devices
 Is there any disadvantage of using parallelization method?
 Memory usage issue

ParallelProgrammingBasics_v2.pdf

Recommended

More Related Content

Similar to ParallelProgrammingBasics_v2.pdf (20)

More from Chen-Hung Hu (12)

Recently uploaded (20)

ParallelProgrammingBasics_v2.pdf