UNIT - IV

1. Illustrate the characteristics of some commobn memory technologies.

Static RAM (SRAM)

• Speed: Very fast (access times in nanoseconds).

• Size: Larger physical size per bit, resulting in lower storage density compared to DRAM.

• Cost: Expensive due to complex circuitry requiring multiple transistors per cell.

• Volatility: Volatile (loses data when power is interrupted).

• Use Case: Primarily used in cache memory (L1, L2) for high-speed data access.

• Advantages: Low power consumption when active, retains state as long as power is applied.

• Disadvantages: High cost and larger size limit its use for large-scale storage.

Dynamic RAM (DRAM)

• Speed: Slower than SRAM, with access times longer due to capacitor-based storage.

• Size: Smaller physical size per bit, enabling higher storage density.

• Cost: Less expensive than SRAM, making it suitable for main memory.

• Volatility: Volatile (requires periodic refreshing to maintain data).

• Types: Includes Asynchronous DRAM, Synchronous DRAM (SDRAM), and Double Data Rate
SDRAM (DDR).

• Use Case: Main memory in computers due to cost-effectiveness and high density.

• Advantages: High capacity and lower cost per bit make it ideal for large applications.

• Disadvantages: Requires refreshing, which adds latency and complexity.

Read-Only Memory (ROM)

• Speed: Moderate, primarily used for read operations.

• Size: Varies based on type (PROM, EPROM, EEPROM, Flash).

• Cost: Generally low cost, especially for mass-produced ROMs.

• Volatility: Non-volatile (retains data without power).

• Types: Includes Programmable ROM (PROM), Erasable Programmable ROM (EPROM),

Electrically Erasable ROM (EEPROM), and Flash Memory.

• Use Case: Stores firmware, OS boot instructions, and embedded system data.

• Advantages: Non-volatile, flexible for specific applications, and retains data without power.

• Disadvantages: Slower write speeds and limited write cycles for flash memory.

Magnetic Disk

Page 1 of 14
 • Speed: Much slower than DRAM, with access times dominated by seek time and rotational
latency.

• Size: Small physical size relative to capacity, offering high storage (e.g., 60 GB for a 1-inch
disk).

• Cost: Low cost per bit, ideal for secondary storage.

• Volatility: Non-volatile (retains data without power).

• Use Case: Secondary storage for large data volumes, such as hard drives and SSDs.

• Advantages: High capacity and low cost make it suitable for bulk storage.

• Disadvantages: Slow access due to mechanical components, which can lead to longer wait
times.

Comparison Table

Memory Type Speed Size Cost

SRAM Very Fast Large Expensive

DRAM Slower Small Less Expensive

Magnetic Disk Much Slower Small Low Price

Conclusion: SRAM is ideal for high-speed cache applications but is costly. DRAM balances cost and
capacity for main memory. ROM variants serve non-volatile storage needs, while magnetic disks
provide high-capacity, low-cost secondary storage despite slower access.

2. Comparing paging and segmentation mechanisms for implementing the virtual memory.

Paging

• Definition: Divides memory into fixed-size units called pages (e.g., 2K to 16K bytes).

• Address Translation: Virtual address comprises a virtual page number and an offset; the
Memory Management Unit (MMU) uses a page table to map virtual page numbers to
physical page frames.

• Mechanism: Pages are loaded into main memory as needed, using Direct Memory Access
(DMA) for disk transfers. A Translation Lookaside Buffer (TLB) caches recent page table
entries to speed up translation.

• Advantages: Simplifies memory allocation with fixed-size pages, efficient memory utilization,
and supports demand paging (loading only required pages).

• Disadvantages: Internal fragmentation can occur, and large page tables require significant
storage.

Segmentation

• Definition: Divides memory into variable-size segments based on logical divisions (e.g., code,
data, stack).

Page 2 of 14
 • Address Translation: Virtual address includes a segment number and an offset; a segment
table maps segment numbers to physical memory.

• Mechanism: Segments are loaded as logical units, potentially requiring contiguous memory,
which can complicate allocation.

• Advantages: No internal fragmentation, as segments are sized to fit data; logical structure
improves modularity and sharing; segment-level protection is intuitive.

• Disadvantages: External fragmentation can occur, and complex memory allocation is needed
due to varying segment sizes.

Comparison Table

Aspect Paging Segmentation

Unit Size Fixed-size pages Variable-size segments

Fragmentation Internal External

Address Translation Page table, TLB Segment table

Memory Allocation Simple, non-contiguous Complex, may require contiguous space

Protection/Sharing Page-level, less intuitive Segment-level, aligns with program logic

Conclusion: Paging is simpler and widely used due to its fixed-size units and efficient memory
utilization, while segmentation offers logical organization but is complex and prone to fragmentation.
Many modern systems combine both techniques to leverage their strengths.

3. Discuss any six ways of improving the cache performance.

1. Increase Cache Size

• Description: Larger caches store more blocks, reducing cache misses.

• Impact: Improves hit rate by holding more frequently accessed data.

• Trade-off: Increases cost and access time due to larger hardware.

2. Use Higher Associativity

• Description: Set-associative or fully associative mapping allows flexible block placement.

• Impact: Higher associativity minimizes conflict misses.

• Trade-off: Increases hardware complexity and comparison time.

3. Optimize Replacement Algorithms

• Description: Use algorithms like Least Recently Used (LRU) to replace the least-referenced
block.

• Impact: LRU improves hit rate by retaining frequently used blocks.

• Trade-off: Requires additional hardware (e.g., counters) to track block usage.

Page 3 of 14
4. Implement Write-Back Protocol

• Description: Update only the cache during writes, marking it with a dirty bit.

• Impact: Reduces memory bandwidth usage, enhancing performance for write-intensive

workloads.

• Trade-off: Adds complexity for dirty bit management and potential data inconsistency risks.

5. Use Multi-Level Caches

• Description: Employ primary (L1) and secondary (L2) caches for improved performance.

• Impact: L1 provides fast access, while L2 reduces misses to main memory.

• Trade-off: Increases complexity and cost of cache systems.

6. Optimize Programs for Locality

• Description: Adjust programs to exploit temporal and spatial locality.

• Impact: Increases hit rate by ensuring frequently accessed data is cached.

• Trade-off: Requires software optimization, which may not always be feasible.

Conclusion: These strategies collectively enhance cache hit rates and reduce miss penalties,
improving overall system performance by addressing both hardware and software aspects of cache
design.

4. Explain the Types of Memories in detail.

Random Access Memory (RAM)

• Definition: Allows read and write operations with equal access time for any memory
location.

• Types:

• Static RAM (SRAM):

• Structure: Uses a latch with cross-connected inverters and transistors to

store 1 bit.

• Operation: Retains data as long as power is applied, no refreshing needed.

• Advantages: Fast access, low power consumption when active.

• Disadvantages: Expensive, larger size per bit, volatile.

• Use Case: Cache memory (L1, L2 caches).

• Dynamic RAM (DRAM):

• Structure: Stores data as a charge on a capacitor, accessed via a transistor.

• Operation: Requires periodic refreshing due to capacitor discharge.

Page 4 of 14
 • Types: Asynchronous DRAM, Synchronous DRAM (SDRAM), Double Data
Rate SDRAM (DDR-SDRAM).

• Advantages: High density, lower cost per bit.

• Disadvantages: Slower than SRAM, requires refreshing, volatile.

• Use Case: Main memory in computers.

Read-Only Memory (ROM)

• Definition: Non-volatile memory primarily used for read operations, retaining data without
power.

• Types:

• Programmable ROM (PROM): User-programmable by burning fuses, irreversible.

• Erasable Programmable ROM (EPROM): Erasable with UV light, reprogrammable.

• Electrically Erasable ROM (EEPROM): Electrically erasable and programmable,

allows selective erasing.

• Flash Memory: High-density, block-based writing, used in flash cards and drives.

• Use Case: Firmware, OS boot code, embedded systems.

• Advantages: Non-volatile, flexible for specific applications.

• Disadvantages: Slower write speeds, limited write cycles for flash memory.

Comparison

• RAM: Volatile, fast, used for temporary storage (cache, main memory).

• ROM: Non-volatile, slower writes, used for permanent storage (firmware, boot code).

• Flash Memory: Bridges ROM and storage, offering high density and re programmability.

5. Write About Memory management requirements.

Address Translation

• Requirement: Translate virtual addresses generated by the processor to physical addresses

in memory.

• Mechanism: The Memory Management Unit (MMU) uses page tables or segment tables to
map virtual addresses to physical locations, supported by a Translation Lookaside Buffer
(TLB) for faster translation.

• Purpose: Ensures programs operate in a virtual address space, independent of physical

memory layout.

Memory Allocation

• Requirement: Allocate memory to processes efficiently, minimizing fragmentation.

Page 5 of 14
 • Mechanism: Paging divides memory into fixed-size pages, allowing non-contiguous
allocation. Segmentation uses variable-size segments, requiring careful management to
avoid external fragmentation.

• Purpose: Maximizes memory utilization and supports multiple processes.

Protection

• Requirement: Prevent unauthorized access to memory regions belonging to other processes

or the OS.

• Mechanism: Page tables include control bits to specify access privileges (e.g., read-only,
read-write). The OS operates in supervisor mode to restrict user programs from accessing
system space.

• Purpose: Ensures data integrity and system security.

Efficient Data Transfer

• Requirement: Minimize delays in transferring data between memory, cache, and secondary
storage.

• Mechanism: Cache uses locality of reference to store frequently accessed data. Virtual
memory uses demand paging and DMA to transfer pages from disk to memory. Interleaving
allows parallel access to multiple memory modules.

• Purpose: Reduces access latency and improves system performance.

Memory Hierarchy Management

• Requirement: Manage a hierarchy of memory types (registers, cache, main memory,

secondary storage) to balance speed, size, and cost.

• Mechanism: Cache stores active data, main memory holds running programs, and secondary
storage provides bulk storage. Multi-level caches (L1, L2) optimize access speed.

• Purpose: Optimizes performance by placing frequently used data in faster memory.

Error Handling and Reliability

• Requirement: Detect and correct memory errors to ensure data integrity.

• Mechanism: Disk controllers use Error Correcting Codes (ECC) to detect and fix errors during
data transfers. Cache coherence protocols ensure consistency across multiple caches.

• Purpose: Maintains reliability in memory operations.

Conclusion: An effective memory management system must provide efficient address translation,
allocation, protection, data transfer, hierarchy management, and error handling to support
multitasking, security, and high performance.

6. Discuss the different mapping techniques used in cache memories and their relative merits and
demerits.

Direct Mapping

Page 6 of 14
 • Description: Each main memory block maps to a specific cache block based on a modulo
operation.

• Address Structure: Divided into tag, block, and word fields.

• Operation: The tag field of the memory address is compared with the tag stored in the cache
block.

• Merits: Simple to implement, low cost, fast access.

• Demerits: High conflict misses due to multiple blocks mapping to the same cache block.

Associative Mapping

• Description: Any main memory block can be placed in any cache block.

• Address Structure: Consists of a tag field and a word field.

• Operation: The tag is compared with all cache block tags simultaneously.

• Merits: High flexibility, better hit rate.

• Demerits: High complexity, slower access due to multiple comparisons.

Set-Associative Mapping

• Description: A hybrid of direct and associative mapping, where cache blocks are grouped
into sets.

• Address Structure: Includes tag, set, and word fields.

• Operation: The set field determines the target set, and the tag is compared within that set.

• Merits: Balanced flexibility, improved hit rate.

• Demerits: Moderate complexity, potential conflicts within a set.

Comparison Table

Technique Flexibility Complexity Hit Rate Cost

Direct Mapping Low Low Low Low

Associative Mapping High High High High

Set-Associative Mapping Medium Medium Medium-High Medium

Conclusion: Direct mapping is simple but suffers from high conflict misses. Associative mapping
maximizes hit rate but is costly and complex. Set-associative mapping strikes a balance, offering good
performance with reasonable complexity, making it a common choice in modern systems.

7. Explain the virtual memory translation and TLB with necessary diagram.

Virtual Memory Translation

• Concept: Virtual memory divides the address space into fixed-size pages. Each virtual
address is translated to a physical address by the Memory Management Unit (MMU).

Page 7 of 14
 • Virtual Address Structure: Comprises a virtual page number (VPN) and an offset.

• Page Table: Stores mappings of virtual page numbers to physical page frames, including
physical page frame address and control bits (valid, dirty).

• Translation Process:

• MMU extracts VPN and offset from the virtual address.

• VPN is combined with the page table base register to locate the corresponding page
table entry.

• If the valid bit is set, the physical page frame address is retrieved, and the offset is
appended to form the physical address.

• If the valid bit is unset (page fault), the OS loads the page from disk into memory,
updates the page table, and retries the access.

• Page Fault Handling: Occurs when a page is not in memory, leading to OS intervention to
load the page from disk. If memory is full, a page is replaced using a replacement algorithm
(e.g., LRU).

Translation Lookaside Buffer (TLB)

• Concept: A small, fast cache in the MMU that stores recently used page table entries to
speed up translation.

• Structure: Contains entries with virtual page number, corresponding physical page frame,
and control bits.

• Operation:

• MMU checks the TLB for the VPN. If found (TLB hit), the physical address is retrieved
immediately.

• If not found (TLB miss), the MMU accesses the page table, retrieves the entry, and
updates the TLB. The OS invalidates TLB entries when page table contents change
(e.g., during context switches).

Advantages

• Reduces translation latency by avoiding page table accesses for frequently used pages.

• Improves performance in systems with large address spaces.

Disadvantages

• Limited size can lead to TLB misses.

• Requires management to maintain consistency with page tables.

Conclusion: Virtual memory translation and TLB mechanisms are essential for efficient memory
management, allowing programs to utilize more memory than physically available while maintaining
performance through fast address translation.

Page 8 of 14
1. What is DMA? Describe how DMA is used to transfer data from peripherals.

Direct Memory Access (DMA)

Definition: DMA is a technique that enables direct data transfer between a peripheral device and
main memory without continuous processor involvement, improving efficiency for high-speed I/O
operations.

How DMA is Used to Transfer Data from Peripherals:

• DMA Controller: A specialized circuit within the I/O device interface that manages DMA
transfers, performing tasks typically handled by the processor.

• Registers in DMA Controller:

• Starting Address Register: Specifies the memory address for the data transfer.

• Word Count Register: Indicates the number of words to transfer.

• Status and Control Register: Includes flags for transfer direction and completion.

• Operation: The processor configures the DMA controller with the starting address and word
count. The controller then takes control of the bus to transfer data directly between the
peripheral and memory.

• Transfer Modes:

• Cycle Stealing: The DMA controller interleaves transfers with processor memory
access.

• Block Mode: The controller gains exclusive memory access to transfer an entire data
block without interruption.

• Priority Handling: DMA requests are prioritized over regular CPU requests, ensuring
that high-speed peripherals can transfer data efficiently without delays.

• Interrupt Notification: Upon completion of a data transfer, the DMA controller can
generate an interrupt to notify the CPU that the data is ready for processing,
allowing the CPU to resume its tasks seamlessly.

Example: A disk controller uses DMA to transfer data blocks from the disk to memory without CPU
intervention, allowing the CPU to perform other tasks simultaneously.

2. Describe in Detail about IOP Organization.

Input/Output Processor (IOP) Organization

An Input/Output Processor (IOP) is a specialized controller that manages I/O operations, reducing
the main processor's involvement in data transfers. The IOP organization includes I/O interfaces,
controllers, and bus structures that facilitate communication between the processor, memory, and
I/O devices.

Components of IOP Organization:

1. I/O Interface:

Page 9 of 14
 • Address Decoder: Identifies the device’s address on the bus for selective
communication.

• Data Register:

• DATAIN: Stores input data from devices.

• DATAOUT: Holds output data for devices.

• Status Register: Tracks the device’s operational status (e.g., ready, busy).

• Control Logic: Manages timing signals and data transfer protocols.

2. DMA Controller:

• Manages direct memory transfers for high-speed devices.

• Supports efficient data movement through cycle stealing or burst mode transfers.

3. Bus Structure:

• A single bus connects the processor, memory, and I/O devices, comprising:

• Address Lines: Specify device or memory locations.

• Data Lines: Transfer data.

• Control Lines: Indicate operations (e.g., read/write).

Mechanisms in IOP Organization:

1. Program-Controlled I/O: The processor polls the device’s status, which can waste CPU time.

2. Interrupt-Driven I/O: Devices send interrupt requests (INTR) when ready, allowing the CPU
to perform other tasks.

3. Direct Memory Access (DMA): The DMA controller manages high-speed transfers without
CPU involvement.

4. Bus Protocols:

• Synchronous Bus: Uses a common clock for data transfers.

• Asynchronous Bus: Uses handshake signals for flexible timing.

Standard I/O Interfaces:

• PCI (Peripheral Component Interconnect): A high-speed expansion bus with multiple

address spaces.

• SCSI (Small Computer System Interface): A high-speed parallel bus for devices like disks.

• USB (Universal Serial Bus): A serial bus that enables plug-and-play for various devices.

Example: In a SCSI disk system, the SCSI controller manages bus control and transfers data to
memory using DMA, notifying the processor via an interrupt upon completion.

3. Describe in Detail about IOP Organization.

Page 10 of 14
Note: This question repeats Question 2. Since no additional unique content is specified for "IOP
Organization," the answer above covers the topic comprehensively. To avoid redundancy, no separate
answer is provided. If a different aspect is intended, please clarify.

4. Discuss the Design of a Typical Input or Output Interface.

I/O Interface Design

Introduction

An I/O interface is a crucial component that connects input/output devices to a computer system,
enabling data transfer between devices and the processor or memory. A well-designed I/O interface
ensures efficient and reliable communication between the system and its peripherals.

Components

1. Address Decoding Circuitry:

• Decodes bus addresses to target the interface.

2. Data Registers:

• Buffers data during transfers (DATAIN for input, DATAOUT for output).

3. Status Registers:

• Indicates device status (e.g., buffer full/empty).

4. Control Logic:

• Manages timing and handshake signals.

5. Data Path (Port):

• Connects the interface to the I/O device (parallel or serial).

Design Considerations

• Bus Protocol Compatibility:

• Aligns with bus timing and control signals.

• Data Format Conversion:

• Converts data formats (e.g., parallel to serial).

• Buffering:

• Ensures stable data transfers.

• Status Monitoring:

• Verifies device readiness.

Examples

• Keyboard Input Interface:

Page 11 of 14
 • Includes an encoder circuit, debouncing circuit, data register, and status flag.

• Display Output Interface:

• Utilizes a data register and control logic to manage data transfer to the display.

5. What are Interrupts? How are they Handled?

What are Interrupts?

Definition: An interrupt is a hardware signal from an I/O device to the processor, indicating that the
device is ready for data transfer or has completed an operation. Interrupts allow the processor to
perform other tasks instead of continuously checking the device.

Handling Steps:

1. Interrupt Request (INTR):

o Device sends an INTR signal.

o Multiple devices share a single line; requests are logically ORed.

2. Processor Response:

o Completes the current instruction.

o Saves the Program Counter (PC) and Status Register to resume later.

3. Interrupt Acknowledgment (INTA):

o Processor sends INTA to confirm receipt.

4. Execute ISR:

o Processor runs the Interrupt Service Routine (ISR) to handle the request.

5. Restore State:

o Reloads saved PC and Status Register.

o Resumes the interrupted program.

Interrupt Control:

• Interrupts are disabled during ISR to prevent conflicts.

• Re-enabled after ISR completion.

Interrupt Latency:

• Delay between INTR and ISR start. Minimized for real-time systems.

Example:
A printer interrupts the CPU when ready for data. The CPU pauses a task, sends data via ISR, then
resumes the task.

Page 12 of 14
6. Give Comparison between Memory-Mapped I/O and I/O-Mapped I/O.

Comparison between Memory-Mapped I/O and I/O-Mapped I/O

Memory-Mapped I/O and I/O-Mapped I/O are two methods for accessing I/O devices, each with
distinct characteristics.

Aspect Memory-Mapped I/O I/O-Mapped I/O

Memory and I/O devices share a common Memory and I/O devices use
Address Space
address space. separate address spaces.

Standard memory instructions (e.g., Move,

Special I/O instructions (e.g., IN,
Instructions Load) handle data transfer. Example: Move
OUT) are required for data
Used DATAIN, R0 transfers keyboard data to register
transfer.
R0.

Uses fewer address lines due to a

Requires more address lines to distinguish
Address Lines dedicated I/O address space,
between memory and I/O.
simplifying hardware.

More flexible, as any memory instruction can Less flexible, limited to specialized
Flexibility
access I/O devices, easing programming. I/O instructions.

Hardware Needs complex address decoding to separate Simpler address decoding due to
Complexity memory and I/O. distinct I/O addresses.

Potentially faster for I/O due to

May be slower due to shared address space and
Performance dedicated addressing and
potential memory conflicts.
instructions.

Uses IN/OUT instructions for

Example Keyboard buffer accessed via Move DATAIN, R0.
separate I/O space.

Suited for systems with distinct

Ideal for systems treating I/O as memory (e.g.,
Use Case I/O handling (e.g., older x86
embedded systems).
architectures).

Key Advantage of I/O-Mapped I/O:

• Reduces interface hardware complexity with fewer address lines.

Key Advantage of Memory-Mapped I/O:

• Simplifies programming by leveraging standard memory instructions, as seen in the keyboard

example.

7. Explain the Action Carried Out by the Processor After Occurrence of an Interrupt

Actions Carried Out by the Processor After an Interrupt

Page 13 of 14
When an interrupt occurs, the processor follows a systematic sequence to handle it while preserving
the state of the current program:

1. Complete Current Instruction: Ensures the ongoing instruction finishes to avoid

inconsistency.

2. Save Program Counter (PC): Stores the address of the next instruction for later resumption.

3. Save Status Register: Preserves the processor’s state (e.g., flags, mode).

4. Disable Interrupts: Prevents new interrupts from disrupting the handling process.

5. Acknowledge Interrupt: Sends an INTA signal to the interrupting device.

6. Load ISR Address: Retrieves the address of the Interrupt Service Routine (ISR) from an
interrupt vector table.

7. Execute ISR: Runs the ISR to address the interrupt (e.g., handle an I/O event).

8. Restore Processor State: Reloads the saved PC and Status Register.

9. Enable Interrupts: Re-allows interrupts.

10. Resume Interrupted Program: Continues the original program from the next instruction.

Example Explanation

In the example, a computation routine is interrupted (e.g., by an I/O device). The processor saves the
PC, executes the ISR to handle the I/O operation, and then resumes the computation at the next
instruction, ensuring seamless program execution.

Interrupt Latency

• Definition: Interrupt latency is the time from when an interrupt occurs to when the ISR
begins executing. It includes time to save registers and context-switch.

• Performance Impact: High latency can slow down system responsiveness, especially if
interrupts are frequent.

• Real-Time Systems: In applications like robotics or medical devices, low latency is critical to
meet strict timing requirements. Excessive latency can lead to missed deadlines or system
failures.

Page 14 of 14