PulseAugur
EN
LIVE 08:52:00

NANQ framework boosts analog compute-in-memory for AI models

By PulseAugur Editorial · [1 sources] ·

Researchers have developed NANQ, a novel quantization framework designed to improve the efficiency and accuracy of analog compute-in-memory (CIM) systems for neural networks. Unlike previous methods that focus on ideal quantization error, NANQ accounts for hardware noise and device variations inherent in CIM arrays. By modeling noise profiles, NANQ adaptively allocates finer resolution to less noisy regions and identifies optimal layer-wise bit-widths, leading to significant accuracy improvements and reduced perplexity in vision and language models. AI

IMPACT NANQ's noise-aware approach could enable more energy-efficient and accurate AI inference on specialized hardware.

RANK_REASON The cluster contains a research paper detailing a new technical framework for AI hardware. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

NANQ framework boosts analog compute-in-memory for AI models

COVERAGE [1]

  1. arXiv cs.AI TIER_1 English(EN) · Yizhe Chen, Wenshuai Yao, Saiya Wang, Yuannuo Feng, Wenbo Qi, Kechao Tang, Ngai Wong, Wenyong Zhou, Wang Kang ·

    NANQ: Noise-Floor-Aware Mixed-Precision Non-Uniform Quantization for Analog Compute-in-Memory

    arXiv:2608.02700v1 Announce Type: cross Abstract: Analog compute-in-memory (CIM) enables energy-efficient neural network inference, but device variation and read noise can severely degrade low-bit quantized models. Existing CIM-oriented quantization methods mainly minimize ideal …

RELATED TOPICS