cpu-scheduling scheduling-algorithms hardware verilog

Combinational Task Scheduling module

A hardware task scheduler design.

Parts of the contribution in FCFS model and SRJF module is made by Eric Chun-Yu

Scheduling Algorithm Module I/O ports

module Scheduling_Algorithm( input clk,
                             input rst,
                             input st,
                             input inputtask,
                             input[20-1 : 0] task_in,
                             output empty,
                             output reg [16-1 : 0] task_out
                           );

I/O	Signal name	Signal definition
input	clk	clock signal
input	rst	synchronous resert signal
input	st	signal to start the module
input	inputtask	raised when there's a task comming into the scheduler
output	empty	fired when no task under schedule
output	task_out	output the scheduled task under current cycle

Scheduling Algorithms

1.First-Come, First Served

simplest scheduling algorithm

Abbreviated as FCFS
Task that request first gets allocated first
Usually comes with a long waiting time
May lead to convoy effect - all tasks wait for a big one to get off

	Arrival time	Burst time	Payload
Task 1	0	7	16’d1
Task 2	2	4	16’d2
Task 3	4	1	16’d3
Task 4	5	4	16’d4

2.Shortest Remaining Job First

preemptive version of Shortest Job First(SJF)

Abbreviated as SRJF
Prioritize the task with shortest remaining time
Advantageous because short processes are handled very quickly
Has the potential for starvation - high priority task keep executing and low priority ones get blocked for indefinite time

3.Shortest Job First

a non-preemptive approach

Abbreviated as SJF
Task having the smallest execution time is chosen for the next execution
May cause very long turnaround times
Also has the potential for starvation

4.Round Robin

everyone gets to play

Abbreviated as RR
Each task is assigned a fixed time slot in a cyclic way
Doesn’t face the issues of convoy effect or starvation.
Depends on the selection of time quantum.
- If too big, it degrades into FCFS.
- If too small, overhead of content switching is too high