Stella Nera: A Differentiable Maddness-Based Hardware Accelerator for Efficient Approximate Matrix Multiplication

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Schönleber, Jannis, Cavigelli, Lukas, Perotti, Matteo, Benini, Luca, Andri, Renzo
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912500287537152
author Schönleber, Jannis
Cavigelli, Lukas
Perotti, Matteo
Benini, Luca
Andri, Renzo
author_facet Schönleber, Jannis
Cavigelli, Lukas
Perotti, Matteo
Benini, Luca
Andri, Renzo
contents Artificial intelligence has surged in recent years, with advancements in machine learning rapidly impacting nearly every area of life. However, the growing complexity of these models has far outpaced advancements in available hardware accelerators, leading to significant computational and energy demands, primarily due to matrix multiplications, which dominate the compute workload. Maddness (i.e., Multiply-ADDitioN-lESS) presents a hash-based version of product quantization, which renders matrix multiplications into lookups and additions, eliminating the need for multipliers entirely. We present Stella Nera, the first Maddness-based accelerator achieving an energy efficiency of 161 TOp/s/W@0.55V, 25x better than conventional MatMul accelerators due to its small components and reduced computational complexity. We further enhance Maddness with a differentiable approximation, allowing for gradient-based fine-tuning and achieving an end-to-end performance of 92.5% Top-1 accuracy on CIFAR-10.
format Preprint
id arxiv_https___arxiv_org_abs_2311_10207
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Stella Nera: A Differentiable Maddness-Based Hardware Accelerator for Efficient Approximate Matrix Multiplication
Schönleber, Jannis
Cavigelli, Lukas
Perotti, Matteo
Benini, Luca
Andri, Renzo
Hardware Architecture
Computer Vision and Pattern Recognition
Machine Learning
Artificial intelligence has surged in recent years, with advancements in machine learning rapidly impacting nearly every area of life. However, the growing complexity of these models has far outpaced advancements in available hardware accelerators, leading to significant computational and energy demands, primarily due to matrix multiplications, which dominate the compute workload. Maddness (i.e., Multiply-ADDitioN-lESS) presents a hash-based version of product quantization, which renders matrix multiplications into lookups and additions, eliminating the need for multipliers entirely. We present Stella Nera, the first Maddness-based accelerator achieving an energy efficiency of 161 TOp/s/W@0.55V, 25x better than conventional MatMul accelerators due to its small components and reduced computational complexity. We further enhance Maddness with a differentiable approximation, allowing for gradient-based fine-tuning and achieving an end-to-end performance of 92.5% Top-1 accuracy on CIFAR-10.
title Stella Nera: A Differentiable Maddness-Based Hardware Accelerator for Efficient Approximate Matrix Multiplication
topic Hardware Architecture
Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2311.10207