mlsanityData to 13 Oct 2026

GRAU: Generic Reconfigurable Activation Unit Design for Neural Network Hardware Accelerators

arXiv 2602.22352 ↗First posted 25 Feb 2026cs.AR cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

With the continuous growth of neural network scales, low-precision quantization is widely used in edge accelerators. Classic multi-threshold activation hardware requires 2n thresholds for n-bit outputs, causing a rapid increase in hardware cost as precision increases. We propose a reconfigurable activation hardware, GRAU, based on piecewise linear fitting, where the segment slopes are approximated by powers of two. Our design requires only basic comparators and 1-bit right shifters, supporting mixed-precision quantization and nonlinear functions such as SiLU. Compared with multi-threshold activators, GRAU reduces LUT consumption by over 90%, achieving higher hardware efficiency, flexibility, and scalability. The best trade-off is usually achieved with 6-8 segments, while complex nonlinearities under aggressive low-cost settings may suffer larger accuracy degradation.

Key facts. No citing papers in the map yet (data through 13 Oct 2026). Builds on 6 of its 22 references that are in the map. First posted on arXiv 25 Feb 2026.

Builds on

6 of 22 references.

References · 6

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,950
  2. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,321
  3. 2017Sigmoid-Weighted Linear Units for Neural Network Function Approximation in Reinforcement Learningcited by 939
  4. 2016FINN: A Framework for Fast, Scalable Binarized Neural Network Inferencecited by 150
  5. 2019Additive Powers-of-Two Quantization: An Efficient Non-uniform Discretization for Neural Networkscited by 148
  6. 2022Power-of-Two Quantization for Low Bitwidth and Hardware Compliant Neural Networkscited by 26

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Optimizing Neural Networks with Learnable Non-Linear Activation Functions via Lookup-Based FPGA Acceleration
  2. 2025A Flexible Precision Scaling Deep Neural Network Accelerator with Efficient Weight Combination
  3. 2026Bitwise Systolic Array Architecture for Runtime-Reconfigurable Multi-precision Quantized Multiplication on Hardware Accelerators
  4. 2022BrainTTA: A 35 fJ/op Compiler Programmable Mixed-Precision Transport-Triggered NN SoC
  5. 2022Low- and Mixed-Precision Inference Accelerators
  6. 2025TYTAN: Taylor-series based Non-Linear Activation Engine for Deep Learning Accelerators ↗
  7. 2023MEGA: A Memory-Efficient GNN Accelerator Exploiting Degree-Aware Mixed-Precision Quantization
  8. 2023On-Chip Hardware-Aware Quantization for Mixed Precision Neural Networks