# AI and ML Accelerator Survey and Trends

**Authors:** Albert Reuther, Peter Michaleas, Michael Jones, Vijay Gadepally, Siddharth Samsi, Jeremy Kepner  
**arXiv:** [2210.04055](https://arxiv.org/abs/2210.04055)  
**Date:** 2022-10-08  
**Summary page:** https://www.summarizepaper.com/en/paper/2210.04055v1/

## Key points

- The paper titled "AI and ML Accelerator Survey and Trends" provides an updated survey of AI accelerators and processors over the past three years.
- The authors have collected and summarized the current commercial accelerators that have been publicly announced with peak performance and power consumption numbers.
- They have plotted the performance and power values on a scatter graph to analyze trends in various dimensions.
- The authors have included two new trends plots based on accelerator release dates in this year's paper.
- Some neuromorphic, photonic, and memristor-based inference accelerators are highlighted as additional trends.
- The analysis sheds light on the current state of AI accelerators and processors in terms of their performance capabilities and energy efficiency.
- This information is essential for researchers working in this field to make informed decisions about which technology to use for their applications.

## Summary

The paper titled "AI and ML Accelerator Survey and Trends" provides an updated survey of AI accelerators and processors over the past three years. The authors, Albert Reuther, Peter Michaleas, Michael Jones, Vijay Gadepally, Siddharth Samsi, and Jeremy Kepner have collected and summarized the current commercial accelerators that have been publicly announced with peak performance and power consumption numbers. They have plotted the performance and power values on a scatter graph to analyze trends in various dimensions. In addition to discussing observations from the trends on this plot, the authors have included two new trends plots based on accelerator release dates in this year's paper. Furthermore, they have highlighted some neuromorphic, photonic, and memristor-based inference accelerators as additional trends. The authors' analysis sheds light on the current state of AI accelerators and processors in terms of their performance capabilities and energy efficiency. This information is essential for researchers working in this field to make informed decisions about which technology to use for their applications. Overall, this paper serves as a valuable resource for anyone interested in understanding the latest developments in AI acceleration technology.