Chapter Three – Memory Management

Memory Management
 Understanding Memory

 What is Memory?
 Collection of data in a specific format.
 Stores instructions and processes data.
 Comprises a large array/group of words or bytes, each with its own
location.
 Primary Purpose: Execute programs.
 Programs and their data must reside in main memory during
execution.
 CPU fetches instructions from memory based on the Program Counter.
 Memory Management

 The Need for Memory Management

 To achieve a degree of multiprogramming.
 To ensure proper utilization of memory resources.
 Many methods exist, reflecting various approaches.
 Effectiveness of algorithms depends on the situation.
 Memory Management
 What is Main Memory? Main Memory (RAM)
 Central to the operation of a modern computer.
 Large array of words or bytes (hundreds of thousands to billions).
 Repository of rapidly available information.
 Shared by CPU and I/O devices.
 Where programs and information are kept for processor utilization.
 Associated with the processor for fast instruction/information
movement.
 Also known as RAM (Random Access Memory).
 Volatile: Loses data upon power interruption.
 Memory Management
 What is Memory Management?
 Primarily involves the management of main memory.
 In multiprogramming, OS resides in one part, multiple
processes use the rest.
 Task: Subdividing memory among different processes.
 Manages operations between main memory and disk
during process execution.
 Main Aim: Achieve efficient utilization of memory.
 Memory Management
 Why is Memory Management Required?
 Allocate and de-allocate memory before and after process
execution.
 Keep track of used memory space by processes.
 Minimize fragmentation issues.
 Ensure proper utilization of main memory.
 Maintain data integrity during process execution.
 Logical vs. Physical Address Space
 Logical Address Space (Virtual Address):

 Address generated by the CPU.

 Represents the size of the process.
 Can change (relative to base).
 Physical Address Space (Real Address):

 Address seen by the memory unit (loaded into Memory Address Register).

 Set of all physical addresses corresponding to logical addresses.

 Computed by the Memory Management Unit (MMU).

 Always remains constant.

 MMU: Hardware device for run-time mapping from virtual to physical addresses.
 Static vs. Dynamic Loading
 Program Loading Techniques:
 Loader: Program that loads a process into main memory.

 Static Loading:
 Loads the entire program into a fixed address at once.
 Requires more memory space (whole program must fit).

 Dynamic Loading:
 A routine is not loaded until it is called.
 All routines reside on disk in relocatable load format.
 Advantage: Unused routines are never loaded, efficient for large code.
 Size of process is not limited to physical memory size.
 Static vs. Dynamic Loading
 Program Linking Techniques

 Linker: Combines object files (generated by compiler) into a single executable file.

 Static Linking:

 Linker combines all necessary program modules (including library routines) into one
executable.
 No runtime dependency on external libraries.

 Executable file size is larger.

 Dynamic Linking:

 Similar to dynamic loading.

 "Stub" included for each library routine reference.

 Stub checks if routine is in memory; if not, loads it.

 Reduces executable size and allows library updates without recompiling

 Swapping
 Definition: Temporarily moving a process from main memory to secondary memory (disk)
and back.

 Purpose:

 Allows more processes to run than can fit simultaneously in memory.

 Handles higher priority processes: lower priority processes can be "swapped out"
(roll-out) to make room.
 Later, swapped back in (roll-in) to continue execution.

 Mechanism:

 Process moves to swap space (secondary memory).

 Main part of swapping is transfer time.

 Total time is directly proportional to the amount of memory swapped.

 Drawback: Involves slow disk I/O, impacting performance if excessive.

 Memory Management Techniques

 Overview of Techniques
 Methods used by an operating system to efficiently allocate, utilize,
and manage memory resources.
 Ensure smooth program execution and optimal use of system
memory.
 Different approaches based on how memory is divided and allocated.
 Memory Management Techniques
 Monoprogramming (Without Swapping)
 Simplest approach.
 Memory divided into two sections:
 Operating System part
 User Program part
 OS keeps track of the first and last location for user programs.
 OS loaded at the bottom or top (often low memory for interrupt
vectors).
 Data/code sharing is not a concern in a single-process
environment.
 Memory Management Techniques
 Multiprogramming with Fixed Partitions
 Fixed Partitioning (Without Swapping)
 Introduced to support multiprogramming.

 Based on contiguous allocation.

 Characteristics:
 Memory partitioned into a fixed number of regions.

 Each partition is a fixed size block of contiguous memory.

 Each partition can hold exactly one process.

 Example: 5 regions (1 for OS, 4 for user programs).

 Management: OS uses a Partition Table to track status.

 Contiguous Memory Allocation
 Each process is given a single, continuous block of memory.

 All data for a process is stored in adjacent memory locations.

 Simpler to manage initially.

 Methods:
 Fixed Partition Allocation: Memory divided into fixed-sized partitions at
system initialization.
 Dynamic Partition Allocation: Memory divided into variable-sized
partitions based on process size.
 When loading a process, OS decides which free block to allocate if multiple
exist.
 Contiguous Memory Allocation
 Partition Allocation Methods (Placement Algorithms)
 Strategies for allocating free blocks in dynamic partitioning:
 First Fit: Allocates the first hole (free block) in memory that is large
enough to fit the process.
 Best Fit: Allocates the smallest hole that is large enough to fit the
process.
 Worst Fit: Allocates the largest hole available, leaving a large
remaining hole.
 Next Fit: Similar to First Fit, but starts searching from where the
previous search left off.
 Non-Contiguous Memory Allocation
 Process is divided into smaller parts.

 These parts are stored in different, non-adjacent memory locations.

 The entire process does not need to be stored in one continuous block.

 Advantages: Eliminates external fragmentation, improves memory

utilization.
 Key Techniques:
 Paging
 Segmentation
 Memory Fragmentation

 Occurs when processes are loaded and removed, creating small, unusable
"holes" in memory.
 These holes may be too small or non-contiguous to be allocated to new
processes.
 Two main types:
 Internal Fragmentation
 External Fragmentation
 Memory Fragmentation
 Internal Fragmentation
 Definition:

 Occurs when memory blocks are allocated to a process more than its requested size.

 Result:

 Unused space is left over within the allocated block.

 Example (Fixed Partitioning):

 Process P4 (2MB) requests memory.

 Allocated a 3MB block.

 1MB block is wasted internally; cannot be allocated to other processes.

 Memory Fragmentation
 External Fragmentation
 Definition: Occurs when there is enough total free memory, but it is broken
up into many small, non-contiguous blocks.
 Result:
 A new process cannot be loaded even if total free space is sufficient, because no single
continuous block is large enough.

 Example:

 Processes P1, P2, P3 allocated (e.g., 2MB in 3MB block, 4MB in 6MB
block, 7MB in 7MB block).
 Leaves 1MB and 2MB free (non-contiguous).
 New Process P4 (3MB) cannot be assigned, even though 3MB total free
Thank You !!