CSC 505

COMPUTER PERFORMANCE EVALUATION

By
Dr. Ogunbiyi
Department of Computer Science and Info Tech
Bells University of Technology

Contents
Lecture 1..................................................................................................................................................2
General Overview.............................................................................................................................2
Lecture 2..................................................................................................................................................8
 Lecture 1
General Overview
Computer performance is the assessment of how effectively a computer
system operates, often quantified by speed, accuracy, and efficiency in
executing program instructions.
For high computer performance, key factors include short response times,
high throughput, efficient resource utilization, data compression
speed, system availability, and network bandwidth.
Key Factors:
1. Short response time – time between when a query is sent and when
feedback is received.
2. High throughput – the rate of processing work.
3. Low utilization of computing resources like memory.
4. Fast data compression – must be highly compact.
5. High availability of the computing system or application.
6. High bandwidth – transfer rate over a network.

Performance can be measured. It is quantitative in nature. It is also

qualitative as evidenced by the above factors.
In technical terms, computer performance can be measured in relation to
two or more systems. Performance is how well the computer can do a work.

Computer Performance Metrics:

1. Availability
 Definition: Availability is a measure of the reliability and accessibility
of a system. It's often expressed as a percentage indicating how much
time the system is operational compared to the total possible time.
High availability is crucial in critical systems like healthcare databases,
financial systems, or e-commerce platforms, where downtime can
mean significant loss of service, trust, and revenue.
Uptime Total Time
 Formula: Availability= × 100
Total Time
  Example in Practice: A system with 99.9% availability (or "three
nines") experiences about 8.76 hours of downtime per year, while
99.99% availability (or "four nines") would allow just under an hour of
downtime annually.
2. Response Time
 Definition: Response time is the interval between a user’s request and
the system’s initial response. It includes all delays, such as network
transmission, processing, and output delivery. Response time is critical
in user-facing applications like websites and software interfaces, where
long delays can lead to poor user experience. Different systems may
prioritize response time differently based on user expectations and
application requirements.
 Formula: Response Time=Processing Time+Transmission Time +Waiting Time
 Example in Practice: In customer service applications, response
times directly impact customer satisfaction. Low response times are
essential in real-time applications, like online gaming or trading
platforms.
3. Channel Capacity
 Definition: Channel capacity represents the maximum data rate a
communication channel can handle without losing information,
measured in bits per second (bps). It is influenced by factors like
bandwidth and noise levels in the communication channel. This metric
is critical in network performance and affects how much data can be
reliably transmitted over a network link.
 Formula (Shannon-Hartley theorem): C=B log 2 (1+SNR)
where C is channel capacity, B is bandwidth, and SNR is the signal-to-
noise ratio.
 Example in Practice: Channel capacity determines video streaming
quality, as higher capacity allows for higher-definition video to be
streamed without buffering.
4. Latency
 Definition: Latency measures the delay from the initiation of a request
to the beginning of the response. Latency is an important performance
factor in networked systems, as it affects the "feel" of responsiveness.
Low latency is especially important in time-sensitive applications like
online gaming, video conferencing, and financial trading, where even
milliseconds of delay can impact performance.
  Formula:
Latency=Propagation Time+Transmission Time + ProcessingTime +Queuing Delay
 Example in Practice: In cloud-based applications, low latency is
essential to ensure that user actions are promptly processed and
displayed, creating a seamless experience.
5. Completion Time
 Definition: Completion time refers to the entire duration needed to
complete a task from start to finish. It encompasses all delays,
processing, and output stages involved in the task. Completion time is
particularly relevant in batch processing or multi-task systems, where
the total time to process jobs is more important than the time taken
per individual task.
 Formula: CompletionTime=Start Time+ Execution Time+ End Time
 Example in Practice: Completion time is a key metric in batch jobs,
like rendering video files or processing large datasets, where the total
duration affects efficiency and scheduling.
6. Service Time
 Definition: Service time is the active period spent by the system
processing a task, excluding any idle or wait times. This metric helps in
understanding the efficiency of the processing power or resources
dedicated to each task.
 Formula: Service Time=CompletionTime−Waiting Time
 Example in Practice: In customer-facing applications, understanding
the service time can help optimize server performance, as higher
efficiency can be achieved by focusing on reducing the time each
request spends in processing.
7. Bandwidth
 Definition: Bandwidth is the maximum rate of data transfer over a
network. It is distinct from throughput, as bandwidth reflects potential
capacity rather than actual performance. Higher bandwidth allows
larger volumes of data to move quickly, supporting more users or
higher-quality data streams.
Total Data Transferred
 Formula: Bandwidth=
Total Transmission Time
  Example in Practice: For streaming services, high bandwidth is
necessary to provide high-definition streaming to multiple users
simultaneously without buffering.
8. Throughput
 Definition: Throughput is the actual rate at which data is successfully
transmitted or processed. It considers the network's or system's
capacity to process data effectively and is a measure of practical
performance, showing the real productivity of a system rather than just
its theoretical limits.
Number of Tasks Completed
 Formula: Throughput=
Total Time
 Example in Practice: Throughput is critical in data centers where
large volumes of data must be processed efficiently, as higher
throughput leads to greater processing power and improved service
delivery.
9. Relative Efficiency
 Definition: Relative efficiency compares a system’s performance
against a baseline or theoretical maximum. It helps in identifying how
well resources are being utilized and is useful for benchmarking and
optimizing systems. Relative efficiency is key in evaluating system
upgrades or when comparing similar systems.
Actual Performance
 Formula: Relative Efficiency= ×100
Maximum Possible Performance
 Example in Practice: In virtual machines or containerized
environments, comparing relative efficiency helps decide the optimal
configuration and scaling based on available resources.
10. Scalability
 Definition: Scalability refers to a system's ability to handle an
increased load or expand without loss in performance. A scalable
system can accommodate additional users or data by adding resources
(like more CPUs or storage) without degrading its performance.
 Example in Practice: In cloud services, scalability is critical as
workloads can vary greatly. Systems that scale well handle fluctuating
demand efficiently, which is ideal for businesses with seasonal or
variable usage patterns.
11. Performance per Watt (Power Efficiency)
  Definition: This metric shows how much computational work is done
per unit of power consumed, representing energy efficiency.
Performance per watt is critical for mobile devices, data centers, and
any other systems where power consumption affects cost or
sustainability.
Performance
 Formula: Performance per Watt=
Power Consumption
 Example in Practice: Data centers often optimize performance per
watt to reduce operational costs and environmental impact, as higher
efficiency reduces power usage and cooling requirements.
12. Compression Ratio
 Definition: Compression ratio measures how much data is reduced by
compression, showing the efficiency of data storage or transmission.
High compression ratios save space and bandwidth but may increase
processing time due to decompression requirements.
 Formula: Compression Ratio=Original ¿ ¿ Compressed ¿ ¿ ¿ ¿
 Example in Practice: Compression ratios are important in video
streaming, where high ratios enable streaming without excessive
bandwidth use.
13. Instruction Path Length
 Definition: Instruction path length counts the number of instructions
executed to complete a task, reflecting the complexity of a process. In
CPU-bound tasks, shorter instruction paths typically yield faster task
completion. Path length can vary based on compiler optimizations,
code complexity, and processor architecture.

 Formula: Instruction Path Length=∑ (Number of Instructions Executed )

 Example in Practice: In computationally intensive applications like

simulations or machine learning, optimizing instruction path length can
improve execution speed.
14. Speed
 Definition: Speed indicates how quickly a system can process data,
often represented by clock speed in processors or general execution
time for specific tasks. Speed is a foundational metric that influences
all other performance aspects and is generally a target for optimization
in both hardware and software.
 1
 Example Formula (Clock Speed): Speed=
Clock Cycle Time
 Example in Practice: Processor speed is crucial in gaming or real-
time applications, where high speed ensures smooth, responsive
performance.
 Lecture 2
Steps in Performance Evaluation:
1. Clearly define the goals – You can compare the system based on the
design or the dimensioning. Comparison must be based on the load
model, in contrast the
2. For any performance evaluation it is important to know whether the
work is based on
3.