Scribd
Upload
2 views

intro_2

Approximate adders are increasingly vital in VLSI design, offering reduced power consumption, circuit complexity, and improved speed for applications like image processing and machine learning. These adders, including architectures like Approximate Mirror Adders and Error Tolerant Adders, balance accuracy and efficiency, making them suitable for energy-efficient digital systems despite challenges like error accumulation. Research shows that optimized approximate adders can significantly enhance image processing performance while maintaining acceptable quality, proving their relevance in modern computing demands.

Uploaded by

raunsekar
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
2 views

intro_2

Approximate adders are increasingly vital in VLSI design, offering reduced power consumption, circuit complexity, and improved speed for applications like image processing and machine learning. These adders, including architectures like Approximate Mirror Adders and Error Tolerant Adders, balance accuracy and efficiency, making them suitable for energy-efficient digital systems despite challenges like error accumulation. Research shows that optimized approximate adders can significantly enhance image processing performance while maintaining acceptable quality, proving their relevance in modern computing demands.

Uploaded by

raunsekar
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 4

Approximate adders have increased importance in present-day VLSI (Very Large Scale

Integration), giving an enough alternative to conventional adders by minimizing power


usage, decreasing circuit complexity, and improving computational speed. The digital
systems are continuing to evolve, the demand is growing for high-speed, energy-efficient,
and low-power arithmetic circuits. Many arising applications such as image processing,
Machine learning, Internet of Things (IoT), etc. These applications can manage minor error
problems, enabling the designer to minimize computational complexity while involving in
matchable output quality. Given the huge increase in the demand for power-efficient
circuits, here approximate adders have become a major component in modern digital signal
processors (DSPs), and embedded computing platforms, in this the performance is restricted
by power dissipation, area used, and delay.

The need for energy-efficient circuits, and high-performance arithmetic circuits has
increased significantly with upgrading in high-performance computing, deep learning, and
artificial intelligence. In most digital architectures, the efficiency of arithmetic units
accurately affects total system performance. Multipliers, and adders, are the main blocks of
computation, play a vital role in judging the power usage, speed, and area efficiency of
present-day processors. Traditional conventional full adders (CFA) are the once that are
made to give exact results but will be suffering from huge power consumption, in increased
in circuit complexity, and propagation delays. These problems make them inefficient for low-
power, and real-time processing applications. To reduce this problem, researchers have
started to use approximate computing, a unique design pattern that relaxes accuracy
problems to improve energy efficiency, computational speed, and hardware complexity.
Approximate computing can be solved in distinct multiple levels, including software-level
approximation, where computational steps are changed to minimize execution complexity,
and hardware-level approximation, in which logical circuits are optimized by declining gate
count, and transistor usage. In arithmetic circuits, the hardware-level approximation is
extremely efficient, as it allows designers to enhance switching activity, minimize sum
propagation delay, and lower the power dissipation, making it extremely capable for next-
generation VLSI applications. Since modern workloads, such as AL-driven applications, edge
computing, and neural networks processing, depend on rapid but power efficient
procedures, approximate computing has come up as a promising technique for energy aware
digital architectures.

One of the major problems in designing arithmetic circuits is carry propagation delay, which
straightly impacts processing speed, and power efficiency. Traditional adders suffer from
long carry propagation chains, restricting their performance in high-speed computing
environments. approximate adders solves this problem by minimizing logic operations,
which is considerably lowers the computational delay, circuit complexity, and power
consumption. Various approximate full adder architectures have been evolved to improve
speed, energy efficiency, and area. Among these, Approximate mirror adders (AMAs) have
accumulated popularity due to their simplistic transistor design, optimized power usage, and
reduced node capacitances. Similarly, Multiplexer Based Approximate adders (MBAFAs) uses
multiplexing methods to obtain a balance between computational accuracy, and circuit
efficiency. Similarly widely used method is Error Tolerant Adders (ETAs), which uses simplistic
carry propagation mechanisms, making them ideal for image processing, real-time
embedded applications, and video encoding. While these designs raises smaller
inaccuracies, they stay within acceptable error thresholds, ensuring reliable functionality in
low-power digital system. Despite their benefits, designing approximate adder also
introduces issues, such as error accumulation, the compromise between accuracy, and
efficiency, and the need for application specific optimization. Careful design approaches
must be engaged to maintain an optimal balance between performance enhancements, and
computational precision, ensuring approximate adders provide efficient yet reliable outputs.

Image processing is a fundamental aspect of current computing, utilized in applications such


as artificial intelligence, multimedia processing, real-time surveillance, and medical imaging.
It includes performing arithmetic operations on pixel values to upgrade, analyze or compress
image for enhance, and efficient storage. Conventional image processing techniques
depends on exact adders, which, while ensuring precision, consume more power, occupy
more cell area, take more time to process the data, and complicate hardware requirements.
In this research, approximate adder are discovered as a more efficient option, consume low-
power, take very less time to process the data, and simplistic hardware requirements while
maintaining acceptable image quality. Since human eye can ignore minor pixel variations,
approximate computing methods can be used in image addition without significantly
impacting visual perception. In this study, two images are added by applying the exact
adder, and the proposed approximate adder, with the outputs analysed based on visual
clarity, and computation efficiency. The approximate adder is developed by modifying the
bit’s in k-map expression, where the minor logic simplification minimizes the circuit
complexity while assuring negligible deviation in the result image. To assess the affect of
approximation, key evaluation metric such as Peak Signal to Noise Ratio (PSNR), Mean
Square Error (MSE), and Structural Similarity Index (SSIM) are applied to measure image
quality. These metrics helps to find how well approximate adders maintain the image
quality while significantly enhancing processing speed, and energy efficiency. By utilizing
approximate computing methods, this research proves that optimized approximate adders
can improve image processing performance, making them highly suitable for real-time AI
based operations, low-power IoT devices, and energy-efficient embedded system.

TOPIC:A Low-Power Yet High-Speed Configurable Adder for Approximate Computing


This circuit minimizes power usage by 42.7% when it is compared to a conventional carry
look-ahead adder,and also minimizes critical path delay by 56.9% for better performance.It is
an compact design which achieves optimization with only an 14.5% area as it compared to
conventional carry look-ahead adder.And also has better image processing.But there is an
potential loss of accuracy when the power is reduced.There is an complexiy in the design
when it compared to traditional approximate adders due to accuracy-configurable
mechanism

Topic: Energy-Efficient Gaussian Filter for Image Processing using Approximate Adder
Circuits
This paper gives the information about an energy-efficient approach for implementing
Gaussian filters by using approximate adder circuits.It saves 40% energy for 3x3 gaussian
filter and 25% for 5x5 gaussian filter.Reduces hardware area usage by 45%.The approximate
3x3 gaussian filter achieves 16.67% timing improvement, while the 5x5 filter achieves a 5.26
%improvement.The loss of accuracy,approximate computing causes some level of error in
image processing.Choosing approximation levels requires detaile simulation and testung to
maintain results.

Topic: Area Efficient Nearly Accurate Approximate Adder Design

This paper introduces an approximate adder that decreases power and area by above 50%
while lowering delay by eliminating carry propagation. It works efficiently for 8-bit,126-bit,
and 32-bit adders with minor accuracy loss. It will face some precision loss, is fitted for
specific applications, and includes design complexity for optimal performance.

Topic: Design of Low Power, Area Efficient, and High-Speed Approximate Adders for
Inexact Computing
This paper introduces a 32-bit approximate adder minimized for low power consumption,
lower delay, and decline in transistor count, gaining 8% power saving using 45-nm
technology. It is approximate computing, lower output voltage swing, and design complexity,
specific application uses only.

Topic: Design and analysis of majority logic-based Approximate Adders and Multipliers
The proposed ML-based approximate computing design enhances working speed, minimizes
usage of power, and optimizes hardware efficiency, making it suitable for scalabale, error
resilient applications like image and pattern recognition. It contains computational errors,
and minimizing it’s use in precision-critical tasks and requiring careful circuit design for
optimal changes. Moreover, error propagation in large-scale circuits can be impacted in total
reliability.

Topic: Power-efficient, high-PSNR Approximate Full Adder Applied in Error-Resilient


Computations Based on CNTFETs
The proposed CNTFET-based adders is suitable for low power consumption, and ha an
improvement in image quality metrics and performs with high speed. It efficiently improves
arithmetic circuit but has minor computation errors and high power dissipation. Moreover,
its design complexity requires careful transistor optimization, and making
implementation problems.

You might also like