# FPIRM: Floating-point Processing in Racetrack Memories

arXiv: https://arxiv.org/abs/2204.13788 · HTML: https://mlsanity.com/paper/2204.13788/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 28 Apr 2022.

## Built on by (most cited first, 1 of 1)

- [Sustainable AI Processing at the Edge](https://mlsanity.com/paper/2207.01209/) (2022)

## Abstract (arXiv)

Convolutional neural networks (CNN) have become a ubiquitous algorithm with growing applications in mobile and edge settings. We describe a compute-in-memory (CIM) technique called FPIRM using Racetrack Memory (RM) to accelerate CNNs for edge systems. Using transverse read, a technique that can determine the number of '1's multiple adjacent domains, FPIRM can efficiently implement multi-operand bulk-bitwise and addition computations, and two-operand multiplication. We discuss how FPIRM can implement both variable precision integer and floating point arithmetic. This allows both CNN inference and on-device training without expensive data movement to the cloud. Based on these functions we demonstrate implementation of several CNNs with back propagation using RM CIM and compare these to state-of-the-art implementations of CIM inference and training in Field-Programmable Gate Arrays. During training FPIRM improves by 2× the efficiency, by reducing the energy consumption by at least 27% and increasing the throughput by at least 18% against FPGA.