Saved in:
Bibliographic Details
Main Author: Prasad, Rohit
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2511.17235
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912723091062784
author Prasad, Rohit
author_facet Prasad, Rohit
contents The increasing diversity and complexity of transformer workloads at the edge present significant challenges in balancing performance, energy efficiency, and architectural flexibility. This paper introduces NX-CGRA, a programmable hardware accelerator designed to support a range of transformer inference algorithms, including both linear and non-linear functions. Unlike fixed-function accelerators optimized for narrow use cases, NX-CGRA employs a coarse-grained reconfigurable array (CGRA) architecture with software-driven programmability, enabling efficient execution across varied kernel patterns. The architecture is evaluated using representative benchmarks derived from real-world transformer models, demonstrating high overall efficiency and favorable energy-area tradeoffs across different classes of operations. These results indicate the potential of NX-CGRA as a scalable and adaptable hardware solution for edge transformer deployment under constrained power and silicon budgets.
format Preprint
id arxiv_https___arxiv_org_abs_2511_17235
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle NX-CGRA: A Programmable Hardware Accelerator for Core Transformer Algorithms on Edge Devices
Prasad, Rohit
Hardware Architecture
The increasing diversity and complexity of transformer workloads at the edge present significant challenges in balancing performance, energy efficiency, and architectural flexibility. This paper introduces NX-CGRA, a programmable hardware accelerator designed to support a range of transformer inference algorithms, including both linear and non-linear functions. Unlike fixed-function accelerators optimized for narrow use cases, NX-CGRA employs a coarse-grained reconfigurable array (CGRA) architecture with software-driven programmability, enabling efficient execution across varied kernel patterns. The architecture is evaluated using representative benchmarks derived from real-world transformer models, demonstrating high overall efficiency and favorable energy-area tradeoffs across different classes of operations. These results indicate the potential of NX-CGRA as a scalable and adaptable hardware solution for edge transformer deployment under constrained power and silicon budgets.
title NX-CGRA: A Programmable Hardware Accelerator for Core Transformer Algorithms on Edge Devices
topic Hardware Architecture
url https://arxiv.org/abs/2511.17235