Systems Programming: File I/O & Optimization

A collection of high-performance C system utilities developed for the "Operating Systems" course at Osnabrück University. This project demonstrates deep understanding of Linux system calls, memory management, and algorithmic optimization.

🚀 Project Overview

The project consists of two distinct modules focusing on different aspects of systems programming:

1. File Copy Benchmarking: A tool to compare different kernel-level strategies for copying data.
2. High-Performance Parser: A highly optimized character frequency analyzer designed to process large datasets under strict time constraints.

🛠 Module 1: Copyfile (I/O Benchmarking)

This module implements and benchmarks three different methods for copying files on a Linux system to understand the overhead of context switches and memory buffers.

Implemented Strategies

• Buffered I/O: Standard read(2) and write(2) loop using a user-space malloc'd buffer.
• Memory Mapping: Uses mmap(2) to map the source file into virtual memory, allowing direct memory-to-file writes via write(2).
• Zero-Copy: Uses sendfile(2) to transfer data directly between file descriptors within the kernel, avoiding user-space data copying entirely.

Usage

The program measures the execution time (in milliseconds) for each method, including resource allocation overhead.

make
./copyfile <mode> <source_file> <destination_file>

Performance Comparison

---

count_naive

Time: 7049479 us

---

count array[512] * 8byte

Time: 1732285 us

---

count array[512] * 8byte + 1 producer & 1 consumer

Time: 1953109 us

---

count array[512] * 8byte + 1 producer & n consumer

Time: 972716 us

---