CITATION.cff

# This CITATION.cff file was generated with cffinit.
# Visit https://bit.ly/cffinit to generate yours today!

cff-version: 1.2.0
title: >-
  Stella Nera: Achieving 161 TOp/s/W with Multiplier-free
  DNN Acceleration based on Approximate Matrix
  Multiplication
message: "If you use it in your work, please cite or reference :-)"
type: software
authors:
  - given-names: Jannis
    family-names: Schönleber
    affiliation: ETH Zürich
    orcid: "https://orcid.org/0009-0000-2242-5331"
  - given-names: Lukas
    family-names: Cavigelli
    affiliation: Huawei Technologies Zürich Research Center
  - given-names: Renzo
    family-names: Andri
    affiliation: Huawei Technologies Zürich Research Center
  - given-names: Matteo
    family-names: Perotti
    affiliation: ETH Zürich
  - given-names: Luca
    family-names: Benini
    affiliation: ETH Zürich
identifiers:
  - type: url
    value: "https://arxiv.org/abs/2311.10207"
  - type: doi
    value: 10.48550/arXiv.2311.10207
preferred-citation:
  type: article
  title: >-
    Stella Nera: Achieving 161 TOp/s/W with Multiplier-free
    DNN Acceleration based on Approximate Matrix
    Multiplication
  authors:
    - given-names: Jannis
      family-names: Schönleber
      affiliation: ETH Zürich
      orcid: "https://orcid.org/0009-0000-2242-5331"
    - given-names: Lukas
      family-names: Cavigelli
      affiliation: Huawei Technologies Zürich Research Center
    - given-names: Renzo
      family-names: Andri
      affiliation: Huawei Technologies Zürich Research Center
    - given-names: Matteo
      family-names: Perotti
      affiliation: ETH Zürich
    - given-names: Luca
      family-names: Benini
      affiliation: ETH Zürich
  doi: "10.48550/arXiv.2311.10207"
repository-code: "https://github.com/joennlae/halutmatmul"
abstract: >-
  The recent Maddness method approximates Matrix
  Multiplication (MatMul) without the need for
  multiplication by using a hash-based version of product
  quantization (PQ). The hash function is a decision tree,
  allowing for efficient hardware implementation, as
  multiply-accumulate operations are replaced by decision
  tree passes and LUT lookups. Stella Nera is the first
  Maddness accelerator achieving 15x higher area efficiency
  (GMAC/s/mm^2) and 25x higher energy efficiency (TMAC/s/W)
  than direct MatMul accelerators in the same technology. In
  a commercial 14 nm technology and scaled to 3 nm, we
  achieve an energy efficiency of 161 TOp/s/W@0.55V with a
  Top-1 accuracy on CIFAR-10 of over 92.5% using ResNet9.
license: MIT