Revolutionizing Machine Learning: the Emergence and Impact of Google’s TPU Technology

EasyChair Preprint 14431

Abstract

This paper provides an in-depth analysis of Tensor Processing Unit (TPU) technology, a custom-built applicationspecific integrated circuit (ASIC) developed by Google for neural network machine learning. We explore the architecture, performance characteristics, and applications of TPUs in modern AI and machine learning tasks. The paper highlights the significant advantages of TPUs over traditional CPUs and GPUs in terms of processing speed and energy efficiency for specific machine learning workloads

Keyphrases: AI hardware, Google, Machine Learning accelearator, Tensor Processing Unit, deep learning

@booklet{EasyChair:14431,
  author    = {S Kasinadhsarma},
  title     = {Revolutionizing Machine Learning: the Emergence and Impact of Google’s TPU Technology},
  howpublished = {EasyChair Preprint 14431},
  year      = {EasyChair, 2024}}