English
Related papers

Related papers: Rethinking floating point for deep learning

200 papers

Is multiplication really necessary for deep neural networks? Here we propose just adding two IEEE754 floating-point numbers with an integer-add instruction in place of a floating-point multiplication instruction. We show that ResNet can be…

Machine Learning · Computer Science 2020-12-08 Tsuguo Mogami

Modern edge AI workloads demand maximum energy efficiency, motivating the pursuit of analog Compute-in-Memory (CIM) architectures. Simultaneously, the popularity of Large-Language-Models (LLMs) drives the adoption of low-bit floating-point…

Hardware Architecture · Computer Science 2026-02-10 Brian Rojkov , Shubham Ranjan , Derek Wright , Manoj Sachdev

In recent years fused-multiply-add (FMA) units with lower-precision multiplications and higher-precision accumulation have proven useful in machine learning/artificial intelligence applications, most notably in training deep neural networks…

Mathematical Software · Computer Science 2019-04-16 Greg Henry , Ping Tak Peter Tang , Alexander Heinecke

Today, almost all computer systems use IEEE-754 floating point to represent real numbers. Recently, posit was proposed as an alternative to IEEE-754 floating point as it has better accuracy and a larger dynamic range. The configurable…

Hardware Architecture · Computer Science 2021-04-13 Varun Gohil , Sumit Walia , Joycee Mekie , Manu Awasthi

With the increasing size of Deep Neural Network (DNN) models, the high memory space requirements and computational complexity have become an obstacle for efficient DNN implementations. To ease this problem, using reduced-precision…

Machine Learning · Computer Science 2019-09-10 Jinming Lu , Siyuan Lu , Zhisheng Wang , Chao Fang , Jun Lin , Zhongfeng Wang , Li Du

Given the current trend of increasing size and complexity of machine learning architectures, it has become of critical importance to identify new approaches to improve the computational efficiency of model training. In this context, we…

Machine Learning · Computer Science 2022-06-08 Badreddine Noune , Philip Jones , Daniel Justus , Dominic Masters , Carlo Luschi

Deep neural networks are commonly developed and trained in 32-bit floating point format. Significant gains in performance and energy efficiency could be realized by training and inference in numerical formats optimized for deep learning.…

Low precision data representation is important to reduce storage size and memory access for convolutional neural networks (CNNs). Yet, existing methods have two major limitations: (1) requiring re-training to maintain accuracy for deep…

Signal Processing · Electrical Eng. & Systems 2020-03-10 Chen Wu , Mingyu Wang , Xinyuan Chu , Kun Wang , Lei He

Quantization has become a mainstream compression technique for reducing model size, computational requirements, and energy consumption for modern deep neural networks (DNNs). With improved numerical support in recent hardware, including…

Computer Vision and Pattern Recognition · Computer Science 2024-05-02 Jordan Dotzel , Gang Wu , Andrew Li , Muhammad Umar , Yun Ni , Mohamed S. Abdelfattah , Zhiru Zhang , Liqun Cheng , Martin G. Dixon , Norman P. Jouppi , Quoc V. Le , Sheng Li

The logarithmic number system (LNS) is arguably not broadly used due to exponential circuit overheads for summation tables relative to arithmetic precision. Methods to reduce this overhead have been proposed, yet still yield designs with…

Numerical Analysis · Mathematics 2020-05-15 Jeff Johnson

State-of-the-art generic low-precision training algorithms use a mix of 16-bit and 32-bit precision, creating the folklore that 16-bit hardware compute units alone are not enough to maximize model accuracy. As a result, deep learning…

Machine Learning · Computer Science 2021-03-09 Pedram Zamirai , Jian Zhang , Christopher R. Aberger , Christopher De Sa

Post-training quantization (PTQ) is a powerful technique for model compression, reducing the numerical precision in neural networks without additional training overhead. Recent works have investigated adopting 8-bit floating-point…

Computer Vision and Pattern Recognition · Computer Science 2024-07-08 Shivam Aggarwal , Hans Jakob Damsgaard , Alessandro Pappalardo , Giuseppe Franco , Thomas B. Preußer , Michaela Blott , Tulika Mitra

The rapid adoption of low-precision arithmetic in artificial intelligence and edge computing has created a strong demand for energy-efficient and flexible floating-point multiply-accumulate (MAC) units. This paper presents a dual-precision…

Hardware Architecture · Computer Science 2026-04-10 Shubham Kumar , Vijay Pratap Sharma , Vaibhav Neema , Santosh Kumar Vishvakarma

One of the major bottlenecks in high-resolution Earth Observation (EO) space systems is the downlink between the satellite and the ground. Due to hardware limitations, on-board power limitations or ground-station operation costs, there is a…

Machine Learning · Computer Science 2023-11-21 Cédric Gernigon , Silviu-Ioan Filip , Olivier Sentieys , Clément Coggiola , Mickaël Bruno

For successful deployment of deep neural networks on highly--resource-constrained devices (hearing aids, earbuds, wearables), we must simplify the types of operations and the memory/power resources used during inference. Completely avoiding…

Machine Learning · Computer Science 2018-10-01 Shumeet Baluja , David Marwood , Michele Covell , Nick Johnston

We propose a cluster-based quantization method to convert pre-trained full precision weights into ternary weights with minimal impact on the accuracy. In addition, we also constrain the activations to 8-bits thus enabling sub 8-bit full…

Machine Learning · Computer Science 2017-02-02 Naveen Mellempudi , Abhisek Kundu , Dipankar Das , Dheevatsa Mudigere , Bharat Kaul

Multiplication is a core operation in modern neural network (NN) computations, contributing significantly to energy consumption. The linear-complexity multiplication (L-Mul) algorithm is specifically proposed as an approximate…

Hardware Architecture · Computer Science 2024-12-30 Ruiqi Chen , Yangxintong Lyu , Han Bao , Bruno da Silva

This preliminary white paper proposes a novel 8-bit floating-point data format HiFloat8 (abbreviated as HiF8) for deep learning. HiF8 features tapered precision. For normal value encoding, it provides 7 exponent values with 3-bit mantissa,…

The biggest challenge for the deployment of Deep Neural Networks (DNNs) close to the generated data on edge devices is their size, i.e., memory footprint and computational complexity. Both are significantly reduced with quantization. With…

Machine Learning · Computer Science 2022-10-17 Cecilia Latotzke , Batuhan Balim , Tobias Gemmeke

Deep neural networks have enabled progress in a wide variety of applications. Growing the size of the neural network typically results in improved accuracy. As model sizes grow, the memory and compute requirements for training these models…