/CPU-job-scheduling-simulator

Primary LanguageJupyter Notebook

CPU Scheduler

A discrete event simulator to conduct experiments on different processor loads and schedulers, and analyse the results to determine in which situations each scheduling algorithm works most efficiently.

How to run it

  1. chmod +x ./run.sh to give it permission to read and write
  2. ./run.sh <integer> to run it with the number of test cases you want or by default it is 15
  3. Result will be in a juypter notebook named, exectued_report.ipynb

Log

  • Bash Script / python script to run all schedulers in one experiment
    • a script to automatically reproduce the files in output/ from the files in input/ and corresponding parameters.prp files for each experiment
  • A python script to extract all the output.out and plot it with a graph

AIM

To investigate the performance of different CPU scheduling algorithms.

Implementation

Performance Metrics

  • Turnaround time [3/9]
  • Waiting time [3/9]
  • Response time [3/9]

Scheduling Algorithm

  • FCFS
  • Round Robin [8/46]
  • Ideal Shortest Job First [10/46]
  • Multi-level feedback queue with Round Robin [13/46]
  • Shortest Job First using exponential averaging [15/46]

Experiments

  • Experiment 1 [13/45]
  • Scope and objectives well-defined [4/13]
  • Parameters and setup explained [3/13]
  • Results visualised [3/13]
  • Results discussed [3/13]

Experiment 2 [13/45]

  • Scope and objectives well-defined [4/13]
  • Parameters and setup explained [3/13]
  • Results visualised [3/13]
  • Results discussed [3/13]

Experiment 3 [13/45]

  • Scope and objectives well-defined [4/13]
  • Parameters and setup explained [3/13]
  • Results visualised [3/13]
  • Results discussed [3/13]

Parameters

Input_generator

  • numberOfProcesses: number of processes
  • staticPriority: priority for each process
  • meanInterArrival: the interval of arrival time
  • meanCpuBurst: CPU burst time
  • meanIOBurstL: IO burst time -> if it is not 0, a process will need to wait
  • meanNumberBursts: burst in process
  • seed: seed

Scheduler

  • scheduler: FcfsScheduler, RRScheduler, IdealSJFScheduler, FeedbackRRScheduler, SJFScheduler
  • timeLimit: whole simulating running time
  • periodic: if periodic, create duplicate events at multiples of inter-arrival time plus cpu plus io time
  • interruptTime: Time for IO waiting
  • timeQuantum: the time that RRScheduler assign for each process, usually around 20 - 50ms
  • initialBurstEstimate: for SJF
  • alphaBurstEstimate: for SJF

This graph has a much stronger relationship than line chart https://matplotlib.org/stable/gallery/statistics/customized_violin.html#sphx-glr-gallery-statistics-customized-violin-py