Comprehensive Review of Computer Organization and Operating Systems: Assignments 5–8

Comprehensive Review of Computer Organization and Operating Systems: Assignments

5–8

Arizona State University

Summer 2024

CSE 230 – Computer Organization and Assembly Language Programming

Homework 5

2. What are the advantages of using DRAM for main memory?

DRAM has a number of strengths that it possesses that make it the best choice of main

memory in a computer system. It has a high storage density, where a high density of memory

can be packed in a smaller physical space than with SRAM. DRAM is also much cheaper to

manufacture and thus it is economical in terms of supplying system memory in large

quantities. Though the performance of DRAM is slower than that of SRAM, its performance

is not too slow to be utilized as the main memory when it is implemented with the cache

memory to accelerate data retrieval speed. Moreover, the DRAM uses less power per bit

stored and is thus used in the modern computers that demand efficient use of power. It is also

easier to produce in larger capacities that meet the growing needs of the modern computer

systems due to its simple design and scalability.

4. Explain the concept of a memory hierarchy. Why did your authors choose to

represent it as a pyramid?

Memory hierarchy is a hierarchical system of various types of memories in a computer

system organized in terms of speed, cost, and capacity. Top of the list are the most expensive

and fastest memory systems including registers and cache that can store small data sets with

high levels of access. Beneath these are the main memory (DRAM) and the secondary

storage systems such as hard drives and SSDs, which are slower, but have a huge storage

capacity. The hierarchy provides access to the data effectively as the most frequently used

data are stored in the faster levels of memory. The memory hierarchy is also represented as a

pyramid by authors because it graphically illustrates the trade-off between the speed, the size,

and the cost the higher you rise the pyramid, the larger the memory and the lower the cost but

access time is slower.

6. What are the three forms of locality?

These three locality include locality in time, locality in space and locality in order. The

accessibility of the same memory locations multiple times in a brief duration of time is

known as temporal locality, e.g. the reuse of a loop variable, or a loop instruction. Spatial

locality is the concept that when memory location is recorded then adjacent memory

locations will soon follow suit as is typical when process arrays or sequential instructions.

Sequential locality is a particular form of spatial locality which happens when data or

instructions are used in a sequential fashion e.g. reading successive lines of code or data

blocks. The three types of locality are crucial in the design of efficient memory systems since

the three types enable the cache and memory hierarchy to make predictions and prefetches on

data that is likely to enhance the system performance.

8. Which of L1 or L2 cache is faster? Which is smaller? Why is it smaller?

The L2 cache is bigger and slower than an L1. It happens to be the closest to the CPU core

and therefore is capable of giving very fast access to the regularly accessed instructions and

data which in turn saves the processor waiting time. There is however, a size limit to L1

cache, due to the very high speed and much more expensive technology used- typically only

a few kilobytes to a few hundred kilobytes. The L2 cache is, however, a little bit slower yet

much larger (measured in megabytes) and acts as a secondary buffer when the CPU misses

the data in the L1 cache. L1 is smaller as a larger L1 would drastically reduce its access time

and cost to manufacture meaning it is its aim to offer the fastest access available to essential

data needed by the processor.

10. What are the three fields in a direct mapped cache address? How are they used to

access a word located in cache?

A direct-mapped cache address is further subdivided into three fields including, the tag, the

index, and the block offset. The index field indicates which line of the cache (or block) to

search in enabling the system to find the possible location of the desired data. This tag field is

then compared with the tag held in that cache line to see whether data being requested is

actually being held this process is called a cache hit when the tags match and the thing is

called a cache miss when they do not. Lastly, the block offset gives the precise position of

the target word in the block of the cache. These three areas combine to allow the processor to

efficiently and swiftly locate the data within the cache of stored information and strikes a

balance between speed and hardware simplicity in the direct-mapped cache structure.

12. Explain how fully associative cache is different from direct mapped cache.

A full associative cache is different in the manner of placement in the cache as compared

with a direct-mapped cache. In direct-mapped cache there are only one fixed number of

cache lines per memory block which are selected by the address index bits. This

deterministic mapping renders accessibility however easy and quick but may create a serious

contention when many blocks overlap with a single block. These are contrasted by a fully

associative cache which permits any memory block to be stored in any line of the cache. This

flexibility minimizes the misses of conflict largely since data may take any available space.

It, however, presupposes more sophisticated hardware, as all the tags in the cache have to be

searched at the same time to retrieve the match. Although fully associative caches are more

performance, they are costlier and require more time to implement, as the complexity is

incremented with this feature.

14. Direct mapped cache is a special case of set associative cache where the set size is 1.

So fully associative cache is a special case of set associative cache where the set size is _

A set-associative cache with a set size equal to the overall number of cache blocks is known

as fully associative cache. That is, every block is a part of one set and hence any block could

be stored at any location within the cache. This is the removal of the constraint of fixed

placement that exists in direct-mapped or smaller set-associative caches, and provides

maximum block placement flexibility and the least possible conflict misses.

16. Explain the four cache replacement policies presented in this chapter.

There are four popular policies to replace the cache, which are Least Recently Used (LRU),

First-In, First-Out (FIFO), Least Frequently Used (LFU) and Random Replacement. The

LRU policy will eliminate the block that is not accessed since a long time with an assumption

that recently accessed data will be required in the nearest future. FIFO removes the oldest

block in the cache, in the same sequence that the data was loaded onto the cache, and does

not understand how frequently the data was used. LFU finds the block with the least number

of accesses and puts it aside giving preference to data with high frequency of access. As the

name indicates, in Random Replacement a block is randomly selected to be replaced and

thereby requires simple hardware design and sometimes works surprisingly well with the

workload. The policies are in balance through complexity, speed and efficiency as related to

the needs of the system.

18. What is the worst-case cache behavior that can develop using LRU and FIFO cache

replacement policies?

Worst-case behavior for both LRU and FIFO is cache thrashing, where nearly every

reference misses. For LRU, the worst case occurs when the program repeatedly touches a

working set slightly larger than the cache (or associativity); each access evicts the block

you’ll need next, so you get a miss on almost every reference. Example: cycling through 4

distinct blocks with a 3-line cache gives a miss for every access. FIFO can also thrash in the

same way, but it has an additional pathology: Belady’s anomaly, increasing the number of

lines can increase misses for some reference patterns. A classic FIFO example (e.g.

1,2,3,4,1,2,5,1,2,3,4,5) shows more misses with 4 lines than with 3, so FIFO can behave

worse than LRU in pathological cases.

20. Explain how to derive an effective access time formula.

The effective access time (EAT) formula measures the average time required to access

memory, considering both cache hits and misses. To derive it, you start with the idea that

some memory accesses are fast (hits) and others are slow (misses). The formula is:

EAT = (Hit rate × Hit time) + (Miss rate × Miss penalty)

Here, the hit rate is the percentage of memory accesses found in the cache, and the miss rate

is 1 minus the hit rate. Hit time is the time to access data from the cache, while miss penalty

is the extra time needed to fetch data from the next memory level. This weighted average

reflects overall performance and helps designers evaluate how cache size, speed, and

efficiency affect total memory access time.

22. What is a dirty block?

A dirty block is a cache block that has been modified in the cache but not yet written back to

main memory. It differs from a clean block, which matches the data stored in main memory.

When a dirty block is replaced, the updated data must be written back to memory to ensure

data consistency.

24. What is the difference between a virtual memory address and a physical memory

address? Which is larger? Why?

Virtual memory address refers to the address given by a CPU when running a program,

whereas physical memory address refers to the whereabouts in main memory (RAM).

Address mapping involves the translation of the virtual address into a physical address with

the help of the Memory Management Unit (MMU). The virtual address space is bigger since

it enables programs to consume more memory than those that are physically allocated by

means of consequently increasing memory capacity through secondary storage (such as a

hard drive or SSD).

26. Discuss the pros and cons of paging.

Paging has some advantages and disadvantages in memory management. On the positive

side, paging removes the problem of external fragmentation by splitting the memory into

fixed-size blocks called pages, thus making memory usage very efficient. Besides that, it

permits non-contiguous allocation that allows processes to use scattered memory spaces and

hence, makes memory allocation and protection easier. In contrast, paging has a few

disadvantages. For example, it increases overhead because of the management of the page

table and the time for translation. Also, continuous page swapping may degrade performance,

thereby, if the system uses virtual memory heavily, it will be affected by thrashing.

Moreover, there still could be some internal fragmentation in the last page of a process.

Nevertheless, paging is still one of the most powerful methods to compromise between the

issues of flexibility, efficiency, and protection in present-day memory management systems.

28. What causes internal fragmentation?

Internal fragmentation is a situation where the memory blocks that have been allocated are

bigger than the actual amount of data, thus, there is some space left inside those blocks that is

not being used. The reason for such a situation is that memory is most of the time divided

into units of a fixed size, e.g. pages or partitions, and a process hardly fits them exactly.

Consequently, the rest of the unused fragments inside the allocated blocks cannot be given to

other processes, hence, there is an inefficient use of memory.

30. What is a TLB and how does it improve EAT?

A Translation Lookaside Buffer (TLB) is a limited, fast-storage unit that is a part of the

CPU’s memory management unit (MMU) and is designed to keep the most recent virtual-to-

physical address translations. On a memory access by the CPU, a TLB search is done for the

address mapping; if it is located (a TLB hit), the translation is immediate thus the page table

in the main memory which is slower is not accessed. What the TLB does is that it cuts down

on the number of times memory must be accessed for address translation and thus the

Effective Access Time (EAT) is greatly lowered which is a big factor for the overall system

performance to be increased.

32. When would a system ever need to page its page table?

A system must do this if it needs to page its page table in the case when the page table is so

large that it cannot fit wholly in main memory. The case is in systems with large virtual

address spaces or when many processes are running simultaneously, each of which requires

its own page table. When the page table is paged, the operating system can keep the

fragments of it on the secondary storage (such as a hard disk or SSD) and can take the

necessary parts into the memory from time to time, thus allowing the memory to be used

efficiently even in large or heavily multitasked environments.

E2. Suppose a computer using direct mapped cache has 2^32 words of main memory and a

cache of 1024 blocks, where each cache block contains 32 words.

a) How many blocks of main memory are there?

Number of main-memory blocks = (2^32 words) / (2^5 words per block) = 2^(32-5) = 2^27

blocks.

Answer (a): 2^27 main-memory blocks.

b) What is the format of a memory address as seen by the cache?

Given:

• Block size = 32 words = 2^5 ⇒ Word (offset) field = 5 bits

• Cache has 1024 blocks = 2^10 ⇒ Index (block) field = 10 bits

• Total address = 32 bits

Tag = 32 - (10 + 5) = 17 bits

Answer (b): Tag = 17 bits, Block (index) = 10 bits, Word (offset) = 5 bits.

Address format: [17-bit Tag | 10-bit Index | 5-bit Offset].

c) To which cache block will the memory reference 000063FA₁₆ map?

Given address: 000063FA₁₆ = 0x63FA = 25594₁₀

Remove offset (5 bits): shift right by 5 → (0x63FA >> 5)

Take next 10 bits for index: (0x63FA >> 5) & (2^10 - 1)

Result: Index = 799₁₀ = 0x31F₁₆

The address 000063FA₁₆ maps to cache block 799 (hex 0x31F).

E4. Suppose a computer using fully associative cache has 2^24 words of main memory and

a cache of 128 blocks, where each cache block contains 64 words.

Given / Useful facts

Main memory size (word-addressable) = 2^24 words → address length = 24 bits.

Block size = 64 words = 2^6 words → word (offset) field = 6 bits.

Cache has 128 blocks = 2^7 entries.

Cache is fully associative (no index field).

(a) Number of main-memory blocks

Number of blocks = (total words) / (words per block)

= 2^24 / 2^6 = 2^18 = 262,144 blocks.

(b) Address format (sizes of tag and word fields)

1. Word (offset) field = log₂(64) = 6 bits.

2. Total address bits = 24, so tag bits = 24 - 6 = 18 bits.

3. There is no index field for fully associative cache.

Thus, the address format is: [Tag: 18 bits] [Word/Offset: 6 bits].

1. Convert the 24-bit address to binary (pad to 24 bits): 01DB72₁₆ =

000000011101100001110010.

2. Split into tag (top 18 bits) and offset (bottom 6 bits):

- Tag (18 bits) = 000000011101100001 → decimal = 1889 (hex 0x761).

- Offset (6 bits) = 110010 → decimal = 50.

3. Memory block number = address >> 6 = 1889. Word-within-block = 50.

4. Since the cache is fully associative, this memory block can be placed in any of the 128 cache

blocks.

The cache stores the tag = 18-bit value (1889) along with the data for block 1889.

Summary for 01DB72₁₆

1. Memory block number = 1889.

2. Word offset within the block = 50.

3. Tag (stored in cache entry) = 18 bits = binary 000000011101100001 (decimal 1889).

4. It may occupy any of the 128 cache blocks (fully associative).

6. A 2-way set associative cache consists of four sets. Main memory contains 2^K blocks of

eight words each.

a) Show the main memory address format that allows us to map addresses from main

memory to cache. Be sure to include the fields as well as their sizes.

In this system, the main memory contains 2^K blocks, and each block holds eight words, which

equals 2^3 words. Therefore, three bits are needed to specify the word (or offset) within a block.

Since the cache is 2-way set associative and has four sets, there are 2^2 sets in total, meaning

two bits are needed to identify the set index. The total number of block bits in memory is K, so

the tag field will consist of the remaining (K - 2) bits. Consequently, the full memory address,

from the most significant to the least significant bits, consists of the tag field (K − 2 bits), the set

index field (2 bits), and the word or offset field (3 bits). The total address size is therefore (K +

3) bits.

b) Compute the hit ratio for a program that loops 3 times from locations 8 to 51 in main

memory. You may leave the hit ratio in terms of a fraction.

The program accesses word locations from 8 to 51, which gives a total of (51 - 8 + 1) = 44

memory references in each loop. Since the program loops three times, there are a total of 44 × 3

= 132 memory accesses. Each cache block contains eight words, so the corresponding block

numbers accessed range from 8 ÷ 8 = 1 to 51 ÷ 8 = 6, meaning that blocks 1 through 6 are

involved in the program’s execution. Because the cache is 2-way set associative with four sets,

each set can hold two blocks, allowing six different blocks to be distributed across the four sets.

During the first iteration, the cache will experience compulsory misses when each of the six

blocks is first brought in. Once all required blocks are loaded into the cache, the subsequent two

loops will result in cache hits for all accesses, since those blocks will already reside in the cache.

Therefore, the total number of misses will be 6 (all occurring during the first iteration), while the

total number of memory accesses is 132.

The hit ratio can thus be expressed as:

Hit Ratio = (Total References − Misses) / Total References = (132 − 6) / 132 = 126 / 132 = 21 /

22.

Hence, the hit ratio for the program is 21/22.

Homework 6

1. State Amdahl’s Law in your own words.

The Law of Amdahl claims that the maximum speed that can be increased in a system by

increasing a single component is bounded by the fraction of time that that one component is

utilised. It finds that raising the quality of a small portion of a process produces decreasing

marginal benefits, and it is critical to improve optimization by attending to activities with the

longest time durations.

2. What is speedup?

Speedup is a measure of how much faster a system performs when enhanced. It is defined as the

ratio of the time taken to complete a task on the original system to the time taken on the

improved system. Mathematically, it is expressed as: Speedup = Execution time (old) /

Execution time (new). A higher speedup value indicates greater performance improvement,

though in practice, speedup is often limited by the parts of the system that cannot be optimized.

3. What is a protocol, and why is it significant in I/O bus technology?

A protocol is a collection of rules and standards that determine the manner in which data is sent

and received between the devices of a computer system. Protocols are therefore an important

aspect in I/O bus technology since they guarantee effective interaction, coordination and

information security between various hardware devices enabling devices of different

manufacturers to communicate and interact effectively.

4. Name three types of durable storage.

Three types of durable storage are magnetic disks, optical discs, and solid-state drives (SSDs).

Magnetic disks store data using magnetic patterns and are reliable for long-term use. Optical

discs use laser technology for reading and writing. Solid-state drives use flash memory, offering

faster access speeds, lower power consumption, and no moving parts, enhancing durability and

reliability.

5. Describe how programmed I/O differs from interrupt-driven I/O.

Programmed I/O makes the CPU monitor the state of an I/O device continuously to decide when

the device is ready to receive data, which may be a waste of processing time. Conversely,

interrupt-driven I/O gives the device the capacity to signal the CPU when it is available, to

enable the processor to execute other activities effectively.

6. What is polling?

Polling is a method in which the CPU repeatedly checks the status of an I/O device to determine

whether it is ready for data transfer. While it provides simple control over devices, it can be

inefficient because the CPU wastes time continuously checking the device instead of performing

other tasks. Polling is best suited for systems with few devices or when real-time response is not

critical.

7. In what way are address vectors utilized in interrupt-driven I/O?

In interrupt-driven I/O address vectors are employed to determine what service routine is

required to deal with a given interrupt. Upon an interrupt being received, the address vector is

used by the system to find the appropriate interrupt service routine and execute it in order to

make sure that the request of every device is met in as short time as possible and in the most

appropriate manner.

8. How does direct memory access (DMA) work?

Direct Memory Access (DMA) allows data to be transferred directly between memory and an

I/O device without continuous CPU involvement. The CPU initiates the transfer by instructing

the DMA controller, which then manages the data exchange. Once the transfer is complete, the

DMA controller notifies the CPU through an interrupt. This process improves system efficiency

by freeing the CPU for other operations.

9. Define a bus master.

A computer bus master is a hardware component or device that controls the system bus and is

able to make data transfers without involvement of the CPU. It handles inter-machine

communication enabling devices to communicate directly in a manner like between memory and

peripherals. This enhances performance since it decreases the use of CPU in data transfer

operations.

10. Why does DMA require cycle stealing?

DMA requires cycle stealing because it temporarily takes control of the system bus from the

CPU to perform data transfers between memory and I/O devices. During this time, the CPU must

wait for bus access, effectively “stealing” one or more cycles from it. Although this causes brief

interruptions, the overall system performance improves since the CPU is relieved from managing

data transfers directly.

11. What does it mean when I/O is described as “bursty”?

When I/O is described as “bursty,” it means that data transfers occur in short, intense bursts of

activity followed by periods of little or no transfer. Instead of a constant flow, the input and

output operations happen irregularly, often depending on when the device or system has data

ready to send or receive.

12. What does it mean when someone refers to I/O as bursty?

When I/O is referred to as bursty, it means that data transfers occur in short, concentrated bursts

rather than at a constant rate. This happens because devices often gather or process data in

chunks before transmitting it. Bursty I/O patterns can cause irregular workloads, requiring

systems to have buffers and scheduling mechanisms to manage these high-activity periods

efficiently.

13. In what ways is channel I/O similar to DMA?

Channel I/O is similar to Direct Memory Access (DMA) because both methods allow data to be

transferred between memory and peripheral devices without constant CPU involvement. In both

cases, the CPU initiates the operation, but the actual data transfer is handled independently,

improving system efficiency and freeing the processor for other tasks.

14. What is multiplexing?

Multiplexing is the technique of transmitting multiple signals over a single communication

channel or bus to improve efficiency. It allows different data streams to share the same physical

medium without interference. There are various types of multiplexing, such as time-division,

frequency-division, and wavelength-division multiplexing, each suited for different applications

in computer networks and I/O systems.

15. What are the main differences between an asynchronous bus and a synchronous bus?

The main difference between an asynchronous bus and a synchronous bus lies in how they

manage timing. A synchronous bus uses a common clock signal to coordinate data transfers,

making communication faster but less flexible. An asynchronous bus, however, transfers data

using handshaking signals instead of a shared clock, allowing devices of different speeds to

communicate reliably.

16. What is settle time, and what can be done about it?

Settle time refers to the delay required for a system, such as a disk drive’s read/write head, to

stabilize over the correct position before reading or writing data. Excessive settle time can slow

performance. To reduce it, manufacturers use better head-positioning mechanisms, faster

actuators, and optimized control algorithms that allow quicker and more accurate head

placement.

17. Why are magnetic disks referred to as direct access devices?

Magnetic disks are called direct access devices because the read/write head can move directly to

the location of the required data instead of reading sequentially from the beginning. This allows

the system to quickly access any specific block of data on the disk, making it much faster than

sequential storage devices like magnetic tapes.

18. Explain the relationship among disk platters, tracks, sectors, and clusters.

A magnetic disk consists of circular platters coated with magnetic material. Each platter surface

contains concentric circles called tracks, which are divided into smaller units called sectors, each

storing a fixed amount of data. The operating system groups one or more sectors into clusters,

which serve as the smallest unit of storage for file management. This hierarchy organizes data for

efficient reading and writing.

19. Explain how disk platters, tracks, sectors, and clusters are related.

Magnificent disk Magnetic disk is a type of magnetic data storage on one or more circular disk

platters coated with a magnetic substance. The platter is further divided into lots of concentric

circles (called tracks) that systemize the data available in the platter. These tracks are again

broken into smaller units known as sectors where each of the sectors holds a set data, whose

average data is 512 bytes or more. In order to manage storage better and efficiently, the operating

systems cluster together a number of sectors and create clusters which are the smallest

addressable unit of file storage. The connection between platters, tracks, sectors and clusters

guarantees effective data structuring and retrieval.

20. What are the major physical components of a rigid disk drive?

The major components of a rigid disk drive include platters, spindle, read/write heads, actuator

arm, and controller electronics. The platters store data magnetically, while the spindle rotates

them at high speed. The read/write heads, mounted on the actuator arm, move across the platters

to access data. The controller electronics manage communication between the drive and the

computer system.

21. Define seek time.

Seek time is the amount of time it takes for a disk drive’s read/write head to move to the track

where the required data is located. It is one of the main factors affecting disk access speed.

Shorter seek times result in faster data retrieval and improved overall system performance.

22. What is the sum of rotational delay and seek time called?

The sum of rotational delay and seek time is known as access time. Seek time measures how

long it takes for the disk’s read/write head to move to the correct track, while rotational delay is

the time needed for the desired sector to rotate under the head. Access time therefore represents

the total delay before data transfer begins and is a key factor in disk performance.

23. What is a file allocation table (FAT), and where is it located on a floppy disk?

A File Allocation Table (FAT) is a data structure used by the operating system to keep track of

where files are stored on a disk. It records which clusters belong to each file and identifies

unused or damaged areas. On a floppy disk, the FAT is located near the beginning of the disk,

right after the boot sector, and is usually stored in duplicate for reliability.

24. By what order of magnitude does a rigid disk rotate more than a flexible disk?

A rigid disk (hard drive) typically rotates at speeds between 4,000 and 15,000 RPM, while a

flexible disk (floppy disk) rotates at around 300 RPM. This means a rigid disk spins about one

order of magnitude faster—approximately 10 to 50 times more. The higher rotational speed

allows hard drives to access and transfer data much faster than floppy disks.

25. What term is used to describe robotic optical disk library devices?

Robotic optical disk library devices are commonly referred to as jukeboxes. These systems use

robotic mechanisms to automatically load and unload optical disks, such as CDs, DVDs, or Blu-

ray discs, into drives as needed. Jukeboxes are typically used in environments requiring large-

scale data storage, backup, and archival with automated retrieval.

EXERCISES

E1. Your friend has just bought a new personal computer. She tells you that her new

system runs at 1GHz, which makes it over three times faster than her old 300MHz system.

What would you tell her?

I would explain to her that although the new computer’s clock speed of 1GHz is more than three

times higher than her old 300MHz system, this does not necessarily mean it will perform three

times faster. The actual speed of a computer depends on many additional factors besides clock

frequency. These include the processor’s architecture, the number of instructions it can execute

per cycle, cache size, memory speed, bus bandwidth, and the efficiency of the operating system

and software. A newer processor may also have improved design and instruction handling, which

contribute more to performance than just clock speed alone.

E11. Why do you think the term random access device is something of a misnomer for disk

drives?

The term random access device is somewhat misleading when applied to disk drives because,

unlike true random access memory (RAM), disk drives cannot access data instantly. To read or

write data, the disk’s read/write head must physically move to the correct track and wait for the

desired sector to rotate under it. This mechanical movement introduces seek time and rotational

delay, making access times much slower and dependent on data location. Therefore, while disk

drives allow access to any block of data without reading sequentially, the access is not truly

“random” in speed or uniformity due to the physical limitations of the device.

12. Why do differing systems place disk directories in different track locations on the disk?

What are the advantages of using each location that you cited?

Different systems place disk directories in various track locations to optimize performance and

reliability based on design priorities. Some systems store directories near the beginning of the

disk to reduce seek time for frequently accessed files, since the outer tracks offer higher data

transfer rates. Others place directories near the middle of the disk to minimize average seek time

when accessing files spread across the entire surface. A few systems even duplicate directories in

multiple locations to enhance fault tolerance and data recovery. Each placement strategy

balances speed, accessibility, and reliability depending on how the system manages file

operations and overall disk usage patterns.

13. Verify the average latency rate cited in the disk specification of Figure 7.11. Why is the

calculation divided by 2?

The disk in Figure 7.11 has a rotational speed of 4,464 revolutions per minute (RPM). To find

the average latency, we first calculate the time for one full revolution:

Time per revolution = 60 seconds / 4,464 revolutions = 0.01344 seconds = 13.44 milliseconds.

Since, on average, the disk head will wait for half a revolution to reach the desired sector, the

average latency is:

Average latency = 13.44 / 2 = 6.72 milliseconds.

The calculation is divided by 2 because, statistically, the desired data will be halfway around the

disk on average. This matches the 6.72 ms value shown in the specification table.

14. By inspection of the disk specification in Figure 7.11, what can you say about whether

the disk drive uses zoned-bit recording?

By examining the disk specification in Figure 7.11, it appears that the disk drive does not use

zoned-bit recording. The specification lists a single value for “Sectors per Track” as 132, which

suggests that every track on the disk contains the same number of sectors. In a zoned-bit

recording system, the number of sectors per track would vary, outer tracks would contain more

sectors than inner tracks to make efficient use of the disk’s surface area. Since only one uniform

value is provided, it indicates that this disk uses constant angular velocity (CAV) recording

rather than zoned-bit recording (ZBR).

Homework 7

26. What is the acronym for computer output that is written directly to optical media

rather than paper or microfiche?

Computer output to Laser Disc is the acronym COLD, meaning computer output direct to optical

media. COLD systems save vast amounts of data, e.g. reports or documents, on optical discs

rather than printing on paper. The approach offers rapid retrieval, long term storage and less cost

in physical storage of information and does not compromise the readability or integrity of

archived information.

27. Magnetic disks store bytes by changing the polarity of a magnetic medium. How do

optical disks store bytes?

Optical disks store bytes using patterns of pits and lands on their reflective surface. A laser beam

reads these patterns, pits scatter light while lands reflect it, to represent binary data (0s and 1s).

Unlike magnetic storage, optical media use light rather than magnetism, making them resistant to

electromagnetic interference and suitable for long-term data preservation.

28. How is the format of a CD that stores music different from the format of a CD that

stores data? How are the formats alike?

A music CD (CD-DA) and a data CD (CD-ROM) differ in how information and error correction

are handled. CD-DA uses a format optimized for continuous audio playback with minimal

interruption, while CD-ROM includes stronger error detection and correction to ensure accurate

data retrieval. Both formats, however, use the same physical structure of pits and lands and

similar optical technology for reading and writing data.

29. Why are CDs especially useful for long-term data storage?

The CDs are useful in long term storage of data since they are resistant to magnetic fields,

environmental conditions and data corruption that are experienced with other media. They have

the advantage of their optical reading mechanism; hence, little wear is experienced as no

physical contact is made between the disc and the laser. Also, CDs are durable with the ability to

provide data integrity of decades, thus making them appropriate in storing valuable files.

30. Do CDs that store data use recording sessions?

Indeed, data CDs have the ability to utilize recording sessions, which makes it possible for users

to add information step by step in several sessions instead of doing it all at once. Each session

contains its own directory structure, thus permitting the CD to be updated at any time. This

capability, referred to as multi-session recording, is quite handy for backups as well as for that

portion of the data which is saved incrementally and stored on writable CDs like CD-Rs.

31. How do DVDs store so much more data than regular CDs?

DVDs hold more data that CDs mostly due to higher data density. They employ a laser with a

shorter wavelength, which enables smaller pits and more closely spaced tracks. In fact, DVDs

may also have several layers on each side of the disc, thus drastically increasing the capacity.

Besides that, DVDs use advanced encoding techniques and more efficient error correction that

when combined, allow for a much larger storage volume in the same physical size.

32. Name the three methods for recording WORM disks.

The three main methods for recording Write Once, Read Many (WORM) disks are burning pits

using a laser, phase-change recording, and magneto-optical recording. In the first, a laser

permanently alters the reflective surface to encode data. In phase-change recording, heat changes

material states to represent bits. Magneto-optical recording combines heat and magnetic fields to

store information that can be read optically but not erased.

33. Why is magnetic tape a popular storage medium?

Magnetic tape remains a popular storage medium due to its high capacity, low cost per bit, and

durability for long-term archival storage. It is ideal for backups and data archives where

sequential access is acceptable. Modern tape technologies, such as LTO (Linear Tape-Open),

offer fast transfer rates and improved reliability, making tape storage a cost-effective solution for

enterprises managing large datasets.

34. Explain how serpentine recording differs from helical scan recording.

In serpentine recording, data is written in linear tracks along the length of the tape; when the end

is reached, the tape head reverses direction to write on adjacent tracks. Helical scan recording,

used in video and some data tapes, writes data diagonally using a rotating head. While helical

scan achieves higher data density, serpentine recording provides greater reliability and simpler

mechanical design.

35. What are two popular tape formats that use serpentine recording?

DLT (Digital Linear Tape) and LTO (Linear Tape-Open) are two well-known tape technologies

that employ serpentine recording to operate. Essentially, both devices inscribe information in

simple, parallel lines on the optical surface and change the direction of the read/write head at the

end the track. These formats are popular in enterprise scenarios due to the fact that they provide

large storage volume, high data rates and durable storage over time.

36. Which RAID levels offer the best performance?

RAID 0 and RAID 10 provide the best performance among RAID levels. RAID 0 stripes data

across multiple disks, significantly increasing read and write speeds but offering no redundancy.

RAID 10 combines mirroring and striping, providing both high performance and data protection.

However, RAID 10 requires more disks and is more expensive than RAID 0.

37. Which RAID levels offer the best economy while providing adequate redundancy?

RAID 5 offers the best balance between cost, performance, and redundancy. It stripes data and

parity information across all disks, allowing recovery from the failure of one drive without losing

data. RAID 5 provides efficient disk utilization since only one disk’s worth of space is used for

parity, making it a cost-effective solution for many business applications.

38. Which RAID level uses a mirror (shadow) set?

RAID 1 uses a mirror set, in which data is duplicated on two or more disks. Every write

operation is performed on both drives, ensuring that if one disk fails, an exact copy of the data is

available on the other. This provides excellent fault tolerance but doubles storage requirements,

as only half of the total disk space is usable.

39. What are hybrid RAID systems?

Hybrid RAID systems combine features of two or more RAID levels to enhance performance,

redundancy, or capacity. Common examples include RAID 10 (a combination of RAID 1 and

RAID 0) and RAID 50 (RAID 5 plus RAID 0). These systems leverage striping for speed and

mirroring or parity for fault tolerance, offering improved overall performance and data

protection.

Homwork 8

1. What was the main objective of early operating systems as compared to the goals of today’s

systems?

Early operating systems were mainly designed to manage hardware in an efficient manner and to

automate basic tasks such as job sequencing, input/output operations, and resource allocation

with the aim of maximizing hardware utilization. Such systems revolved around the idea of

increasing the computational speed and cutting down the time during which the machine would

be idle in a batch processing environment. Nowadays, operating systems have developed their

functionalities to focus more on factors like user-friendliness, multitasking, security, and

networking as well as distributed computing support.

Contemporary systems strive to deliver an interactive, smooth, and user-friendly experience

while handling complex hardware and software interactions. They focus on attributes such as

reliability, scalability, and real-time responsiveness in order to be able to serve a wide range of

applications which, in turn, include personal computing, large-scale enterprise, and cloud

environments and, thus, demonstrate the transition from efficiency to usability and connectivity.

2. What improvements to computer operations were brought about by resident monitors?

Resident monitors made a major impact on how computers operated by automating and

simplifying program execution, which was a manual task. In a manual operation, the operators

had to load each program, input data, and start execution, which was very time-consuming and

error-prone. Resident monitors came up with the concept of control programs permanently

residing in memory to automatically handle job sequencing. They did work like loading

programs, changing jobs, and handling input/output operations, thus it was a big time and human

help saving. This automation resulted in better CPU utilization as the system could switch from

one job to another without long waiting times. Resident monitors paved the way for more

complex operating systems by bringing in some of the concepts like job control languages,

buffering, and scheduling. These improvements were a major milestone in the direction of

efficient, reliable, and user-independent computing that made multiprogramming and interactive

operating systems of later generations possible.

3. With regard to printer output, how was the word spool derived?

The term "spool" comes from "Simultaneous Peripheral Operations On-Line." It was an idea that

came about in the times of the first computers when printers and other peripherals were very

slow compared to the central processing unit (CPU). If there were no spooling, the CPU would

still be able to wait among other things while a printer does its work, which leads to a waste of

the system resources. The concept of spooling was introduced as a solution to this problem. In

this operation, data going to output devices like printers is temporarily stored in a place - usually

on a disk - where it can be queued and handled without the need of the CPU. Once saved, the

data is fetched and printed at the peripheral's speed while the CPU can still work on other tasks.

This tool allowed overlapping of input, processing, and output operations, thus it was the most

efficient way of using the system. Spooling was one of the first ways of multiprogramming

coming from the past and it is still a basis for modern operating systems in handling print jobs

and other buffered I/O operations.

4. Describe how multiprogramming systems differ from timesharing systems.

Both multiprogramming and timesharing systems are intended to enhance the use of computer

resources, but their differences are quite substantial in terms of objective, architecture, and user

interaction. A multiprogramming system is a typical example of how to raise the efficiency of

the Central Processing Unit by ensuring that several programs can be kept in the main storage at

the same time. The OS selects one program to run while others are at the stage of I/O. When one

program is waiting for I/O, the CPU is changed to another one so that there is as little

consumable time as possible and the process goes continuously. Directly in multiprogramming

environments, users are usually not given access to the system during the run of their programs.

On the other hand, timesharing systems take the concept of multiprogramming further by

allowing many users to interact with the computer at the same time. The CPU's time is

segmented into very short intervals or "time slices," and every user or process is assigned a brief

period of computing time one after another in a very quick manner. As a result, the idea is

generated that each user is the sole possessor of the computer. The primary features of

timesharing are interactivity, low delay, and fairness thus real-time communication between

users, and the system is possible. Whereas in multiprogramming the main focus is on achieving

the maximum utilization of the CPU, timesharing just turns that attention to the efficiency of the

user interaction and response time, thus it is perfect for present multiuser environments like

servers, terminals, and interactive computing platforms.

5. What is the most critical factor in the operation of hard real-time systems?

Perhaps the most important aspect in the functioning of hard real-time systems is the capability

of the system to time its operations accurately and to less than a percent of the time in the case of

chance. It is a situation where it is required that tasks be achieved within a time frame which is

fixed beforehand; just a little delay can cause very severe consequences, for example in instances

of medical instruments, air traffic control, or industrial automation. Hence, the emphasis has

been laid on predictability and determinism rather than putting the system to work at higher

speeds of processing. The operating system thus must make it possible that the most urgent tasks

are carried out without delay, and resource allocation, scheduling, and interrupt handling are

tightly managed. Real-time schedulers are generally employing scheduling algorithms such as

Rate Monotonic or Earliest Deadline First to perform task execution timely. To safeguard the

continuation of a safe and sound operation, system stability and reliability, along with low

latency, and steady response time, are critical since any deadline missed can lead to the

breakdown of the whole system and consequently be fatal.

6. Multiprocessor systems can be classified by the way in which they communicate. How are

they classified in this chapter?

The chapter describes different types of multiprocessor systems. These systems can be either

tightly or loosely coupled systems based on how their processors communicate and share data

during operation. Tightly coupled systems mean that several processors share the same main

memory and one operating system controls the whole system. The processors can access the

shared memory directly so they can exchange data quickly and perform tasks in a coordinated

manner. Load balancing becomes easier, inter-process communication becomes faster, and

control through a single central unit becomes possible with this kind of architecture. That is why

these systems are the ones that scientific computations, large-scale simulations, or other

applications requiring high-speed processing and close synchronization use.

On the other hand, a loosely coupled system, which is also known as a distributed system, is a

collection of independent processors or computers connected by communication links like

networks. Each processor has its own local memory, and they all function under different

operating systems. Coordination is obtained through message passing, not by shared memory.

Such systems become more flexible and scalable because of this structure. Individual nodes can

work independently or they can cooperate. Loosely coupled systems find their use in places

where tasks can be easily split and assigned to different machines. Thus, data centers, clusters, or

cloud computing infrastructures are some examples of environments where these systems are

implemented.

The chapter elaborates that the main difference between these two groups of systems is their way

of communication and degree of integration. Tightly coupled systems have centralized control

and high-speed interconnectivity which make the parallel processing with low latency possible.

On the contrary, loosely coupled systems are characterized by modularity, fault tolerance, and

scalability features. They also suffer from slower communication since there is a network

overhead. Each architecture has its own advantages and disadvantages, and thus which one will

be selected depends on the performance, cost, system complexity, and other criteria of the

application. This classification is the basis for the understanding of today's multi-core processors

and distributed computing environments.

7. How is a distributed operating system different from a networked operating system?

A distributed operating system (DOS) is different from a networked operating system (NOS) in

the sense that it provides a higher degree of integration and transparency to users and

applications. A network operating system connects individual computers through a network, but

each machine still has its own operating system and resources. Users need to log into specific

machines and manually access remote files or resources. The resources of the system are

separate and the network is used as a means of communication between them, hence they are

visible and managed separately.

On the other hand, a distributed operating system organizes a set of linked computers as one

single system. It does not reveal the fact that the resources are distributed in different places but

creates the illusion of one machine for users. The DOS deals with communication, resource

sharing, and load balancing among all nodes without the user having to intervene. Users are

ignorant of the location of programs or data since the system is transparent in access and

management. Such a high level of integration is instrumental in improving efficiency, fault

tolerance, and scalability, as the execution of tasks can be split across multiple processors.

Hence, while a networked operating system is concerned with establishing connections, a

distributed operating system ensures uninterrupted cooperation, resource transparency, and

centralized control in the dispersed environment.

8. What is meant by transparency?

Transparency in computing means that a distributed system can show its resources and

operations as one single entity, without revealing the hardware and network details. Users and

applications can interact with files, processes, or devices in a straightforward manner, without

requiring them to be aware of the physical location or management.

9. Describe the two divergent philosophies concerning operating system kernel design.

The two different views of an operating system kernel are monolithic kernel and microkernel. A

monolithic kernel is one in which all the basic operating system services such as process

management, memory management, device drivers, and file systems are run in kernel mode as

one single large program. This architecture enables very fast communication between the system

components because the calls are done directly, hence, the system performs at a very high level.

Nonetheless, it also elevates the system to be very complex and exposes it to be highly

vulnerable; any other error in one of the modules can lead to a total system crash.

On the other hand, a microkernel attempts to reduce the tasks of the kernel by only including

those that are absolutely necessary such as interprocess communication and basic scheduling.

Other services like device drivers, file systems, and networking are user-level processes. Such a

division helps to improve the overall stability, adaptability, and safety of the system as faults in

one service will not affect the whole system. Although micro-kernels may have a slight

performance drop, they provide better modularity and easier maintenance.

10. What are the benefits and drawbacks to a GUI operating system interface?

A Graphical User Interface (GUI) operating system is packed with a multitude of benefits that

have propelled it to a modern computing standard; nevertheless, it has a few drawbacks as well.

Ease of use is one of the major strengths of GUI. GUI, by its very nature, is based on such visual

elements as icons, windows, buttons, and menus, and it thus enables users to communicate with

the system in an intuitive way, not requiring them to memorize complex command lines.

Consequently, users with a little or no background knowledge of the system and those coming

from non-technical fields are the most appropriate audience for a GUI. Moreover, GUIs allow

users to increase their output by giving them the opportunity to perform several tasks

simultaneously. i.e., users can open, move, and manage several applications at a time. Since GUI

is a visual medium, it also becomes very easy to learn, follow, and do the work very quickly,

thus cutting the training time and creating a friendly user environment.

On the contrary, GUIs are associated with some disadvantages when compared to command-line

or text-based interfaces. Their dependency on elaborate graphics and multiple background

processes results in GUI systems being the major users of computer resources such as memory,

processing power, and storage. Consequently, they are not the right choice for situations where

utmost efficiency is required or for gadgets with hardware of low capacity. Furthermore, while

GUIs make it easy to perform daily tasks, they usually have a lesser degree of user control and

flexibility. In fact, advanced users might encounter limitations in the sense that intricate settings

and system functionalities may be concealed or via graphical tools inaccessible. Additionally, the

speed at which repetitive or batch operations are done in GUIs is low when compared with

command-line interfaces, which can be automated through scripting.

To sum up, GUI operating systems have changed the face of computing by making technology

more user-friendly, interactive, and visually appealing. The use of these systems is highly

recommended in personal computing, education, and business environments where great

importance is attached to usability. Nevertheless, their heavy demand for resources, the limited

control for expert users, and the slower performance in certain tasks serve as pointers that, while

GUIs facilitate user experience, they do so at the expense of efficiency and simplification of the

system.

11. How is long-term process scheduling different from short-term process scheduling?

Long-term and short-term process scheduling mainly differ in their intent, frequency, and effect

on a system's performance. Long-term scheduling or job scheduling is the process that figures

out which programs get the green light to be run in the system. It regulates multiprogramming

level - how many processes can be in memory at the same time - by picking jobs from the job

pool kept on secondary storage and getting them ready in the main memory. Such scheduling is

rare and only concentrates on mixing I/O-bound and CPU-bound processes to achieve the best

system throughput.

On the other hand, short-term scheduling or CPU scheduling is a much more frequent operation

that selects the next process to be executed from the ready queue in memory. The primary

objective here is to keep the CPU busy and to provide very short waiting times especially in

interactive systems. Whereas long-term scheduling is about load and overall system performance

management, short-term scheduling is about immediate CPU allocation ventilating which tasks

get executed next.

12. What is meant by preemptive scheduling?

Preemptive scheduling is a CPU scheduling strategy addressed by an operating system when it

decides to take away a running process from the CPU forcibly and grant the CPU to another

process. The philosophy behind this method is that the CPU should be given to the process with

the highest priority or the shortest calculation time thus ensuring both fast interaction with the

user and system performance. The process in execution can be interrupted in preemptive

scheduling if there is a higher priority process in the queue or if the time given to the running

process has expired.

This scheduling method is still very important in multitasking and real-time systems found in

modern computers which are very sensitive to user actions and time. Preemptive scheduling

algorithms are such as Round Robin, Shortest Remaining Time First (SRTF), and Priority

Scheduling with a preemption feature. The leading edge of preemptive scheduling is that the

CPU can be used efficiently and the system can be responsive. The problem is that there is some

overhead caused by very frequent context switching which is a process of saving the state of the

interrupted process and loading the state of the new one. Being aware of this disadvantage, still,

preemptive scheduling is implemented in modern operating systems like Windows, Linux, and

macOS.

13. Which method of process scheduling is most useful in a timesharing environment?

Round Robin (RR) scheduling is generally regarded as the most efficient and popular method to

be implemented in a time-sharing environment. This kind of scheduling is aimed at preserving

fairness and promptness when multiple processes compete to be given CPU time in a

multitasking system. In the Round Robin scheduling, each process in the ready queue receives a

fixed time slice during which it is allowed to execute. When a process’s time slice comes to an

end, the process is forcibly removed and put at the end of the queue, while the CPU is given to

the next process. This loop goes on until all processes have been run.

What makes Round Robin scheduling so powerful is essentially its capacity to give each process

the same chance to make use of the CPU, thereby, no process is allowed to dominate the system

resources. Thus, it is the best choice for time-sharing systems, where users or tasks are many and

each CPU access is for a short period. Furthermore, it makes possible for interactive processes to

have good response times, e.g., those used in user terminals. It should, however, be mentioned

that the success of this technique is greatly influenced by the duration of the time quantum—i.e.,

if the time is too short, context-switching overhead will be high; if it is too long, the system will

not be very responsive. So, basically, Round Robin is a way to achieve balanced CPU allocation

in shared computing environments.

14. Which process scheduling method is provably optimal?

The Shortest Job Next (SJN) or Shortest Job First (SJF) is the one that is usually referred to as a

‘provable optimal’ type of process scheduling solution when it comes to the average waiting time

of processes in the system. To be more specific, this algorithm selects the process with the lowest

estimated CPU burst time to be executed next. Simply, the one is given priority over others

which is the process with the least amount of CPU time. The reasoning behind such an approach

is straightforward: by doing the shortest processes first, the overall waiting time in the ready

queue of all processes is minimized.

Non-preemptive and preemptive structures are the two basic categories of Shortest Job First

scheduling, with the latter being most commonly referred to as Shortest Remaining Time First

(SRTF). In the non-preemptive version, the process, which results in execution, is run to

completion. When talking about the preemptive version, the interruption of the current process

can be done by a newly arrived process with a shorter remaining burst time. Both variations are

designed to lead to optimal average waiting times if the conditions are ideal.

On the other hand, one can argue that while SJF is an optimal solution in theory, it is far from

being a practical one in reality of any modern computing environment since it demands an

accurate and upfront knowledge about the CPU burst of each process. To be more specific,

operating systems are now required to figure out the next CPU burst by looking at the past

average or by applying exponential averaging techniques, nevertheless, such predictions are

sometimes far away from the truth. Also, SJF can eventually be the cause of starvation for those

processes which are long enough in the case when it goes on this way that there always comes a

short process and thus these long ones may never be given CPU time.

Meanwhile, the Shortest Job First technique can still be regarded as a good instrument to

compare performance of various scheduling strategies besides showing its serious drawbacks.

This particular approach and its proof of optimality constitute one of the core concepts not only

in the design and performance evaluation of the operating system but also in coming up with

next generation of emerging scheduling strategies like multilevel feedback queues that are more

adaptive and equitable.

15. Describe the steps involved in performing a context switch.