tuGEMM: Area-Power-Efficient Temporal Unary GEMM Architecture for Low-Precision Edge AI

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Nair, Harideep, Vellaisamy, Prabhu, Chen, Albert, Finn, Joseph, Li, Anna, Trivedi, Manav, Shen, John Paul
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910761249406976
author Nair, Harideep
Vellaisamy, Prabhu
Chen, Albert
Finn, Joseph
Li, Anna
Trivedi, Manav
Shen, John Paul
author_facet Nair, Harideep
Vellaisamy, Prabhu
Chen, Albert
Finn, Joseph
Li, Anna
Trivedi, Manav
Shen, John Paul
contents General matrix multiplication (GEMM) is a ubiquitous computing kernel/algorithm for data processing in diverse applications, including artificial intelligence (AI) and deep learning (DL). Recent shift towards edge computing has inspired GEMM architectures based on unary computing, which are predominantly stochastic and rate-coded systems. This paper proposes a novel GEMM architecture based on temporal-coding, called tuGEMM, that performs exact computation. We introduce two variants of tuGEMM, serial and parallel, with distinct area/power-latency trade-offs. Post-synthesis Power-Performance-Area (PPA) in 45 nm CMOS are reported for 2-bit, 4-bit, and 8-bit computations. The designs illustrate significant advantages in area-power efficiency over state-of-the-art stochastic unary systems especially at low precisions, e.g. incurring just 0.03 mm^2 and 9 mW for 4 bits, and 0.01 mm^2 and 4 mW for 2 bits. This makes tuGEMM ideal for power constrained mobile and edge devices performing always-on real-time sensory processing.
format Preprint
id arxiv_https___arxiv_org_abs_2412_17966
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle tuGEMM: Area-Power-Efficient Temporal Unary GEMM Architecture for Low-Precision Edge AI
Nair, Harideep
Vellaisamy, Prabhu
Chen, Albert
Finn, Joseph
Li, Anna
Trivedi, Manav
Shen, John Paul
Hardware Architecture
Artificial Intelligence
Machine Learning
General matrix multiplication (GEMM) is a ubiquitous computing kernel/algorithm for data processing in diverse applications, including artificial intelligence (AI) and deep learning (DL). Recent shift towards edge computing has inspired GEMM architectures based on unary computing, which are predominantly stochastic and rate-coded systems. This paper proposes a novel GEMM architecture based on temporal-coding, called tuGEMM, that performs exact computation. We introduce two variants of tuGEMM, serial and parallel, with distinct area/power-latency trade-offs. Post-synthesis Power-Performance-Area (PPA) in 45 nm CMOS are reported for 2-bit, 4-bit, and 8-bit computations. The designs illustrate significant advantages in area-power efficiency over state-of-the-art stochastic unary systems especially at low precisions, e.g. incurring just 0.03 mm^2 and 9 mW for 4 bits, and 0.01 mm^2 and 4 mW for 2 bits. This makes tuGEMM ideal for power constrained mobile and edge devices performing always-on real-time sensory processing.
title tuGEMM: Area-Power-Efficient Temporal Unary GEMM Architecture for Low-Precision Edge AI
topic Hardware Architecture
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2412.17966