UNIVERSITY OF WEST ATTICA
SCHOOL OF ENGINEERING
DEPARTMENT OF COMPUTER ENGINEERING AND INFORMATICS

University of West Attica · Department of Computer Engineering and Informatics

---

Operating Systems II

Parallel Inner Product Calculation using C POSIX Threads

Vasileios Evangelos Athanasiou
Student ID: 19390005

GitHub · LinkedIn

---

Supervision

Supervisor: Vasileios Mamalis, Professor

UNIWA Profile

Co-supervisor: Nikolaos Psarras, Applications Lecturer

UNIWA Profile · LinkedIn

---

Athens, June 2022

---

---

README

Parallel Inner Product Calculation using C POSIX Threads

This repository contains an assignment for Operating Systems II, focusing on parallel computation using POSIX threads in C to efficiently calculate the inner product of two vectors.

---

Table of Contents

Section	Folder/File	Description
1	assign/	Assignment material for the Threads workshop
1.1	assign/ASK-2A-OS-II-LAB-2021-22.png	Assignment description in English
1.2	assign/ΑΣΚ-2Α-ΛΣ-ΙΙ-ΕΡΓ-2021-22.png	Assignment description in Greek
2	src/	Source code demonstrating thread usage
2.1	src/pthreads.c	POSIX threads (pthreads) example implementation
3	README.md	Project documentation
4	INSTALL.md	Usage instructions

---

1. Project Overview

The project implements a C program that calculates the inner product of two vectors using multiple threads.

The inner product is defined as:

A₁ × B₁ + A₂ × B₂ + ... + Aₙ × Bₙ

where Aᵢ and Bᵢ are elements of vectors A and B.

The computation is parallelized by dividing the workload among multiple threads.

---

2. Objectives

• Compute the inner product using p parallel threads.
• Assign each thread a portion of the vector to compute.
• Combine partial sums into a shared global result.
• Protect shared data using mutex synchronization.
• Measure performance with different thread counts (1, 2, 4, and 8).
• Support vector input via user input or file loading.

---

3. Key Features

• Dynamic Thread Creation
  Number of threads is determined at runtime.

• Local Computation
  Each thread computes a local partial sum independently.

• Mutex Synchronization
  Critical sections are protected to prevent race conditions.

• Random Vector Initialization
  Allows performance testing with large datasets.

---

4. Program Structure

1. Read vector size n and thread count p.
2. Allocate memory for vectors A and B.
3. Create threads to compute partial products.
4. Synchronize updates to shared variable total_sum.
5. Output the final inner product result.