arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2602.00838cs.ARcs.AI
收藏

一元算术运算基矩阵乘法单元的探索用于低精度深度学习加速器

Exploration of Unary Arithmetic-Based Matrix Multiply Units for Low Precision DL Accelerators

Department of Electrical and Computer Engineering, University of Central Florida(电气与计算机工程系,中央佛罗里达大学)

Prabhu Vellaisamy, Harideep Nair, Di Wu, Shawn Blanton, John Paul Shen

AI总结:

本文探讨了单元GEMM在低精度深度学习加速器中的应用,通过对比三种最新设计评估其性能与优化点。

摘要:

通用矩阵乘法(GEMM)是深度学习(DL)中的基本操作。随着DL越来越倾向于低精度,近期的研究提出了新的单元GEMM设计作为传统二元GEMM硬件的替代方案。需要严格评估最近的单元和二元GEMM设计,以评估单元硬件在未来DL计算中的潜力。本文聚焦于整数基DL推断的单元GEMM设计,并通过与传统二元GEMM进行比较,对三个最新的单元设计提案,即uGEMM、tuGEMM和tubGEMM进行了详细评估。在不同位宽和矩阵大小下进行了严格的后综合评估,以评估设计的权衡并确定最佳折中点。进一步,我们对八个预训练的卷积神经网络(CNNs)和LLaMA2大语言模型(LLM)进行了权重稀疏性分析。在本工作中,我们展示了如何有效地将单元GEMM用于未来边缘AI加速器中的节能计算。

英文摘要:

General matrix multiplication (GEMM) is a fundamental operation in deep learning (DL). With DL moving increasingly toward low precision, recent works have proposed novel unary GEMM designs as an alternative to conventional binary GEMM hardware. A rigorous evaluation of recent unary and binary GEMM designs is needed to assess the potential of unary hardware for future DL compute. This paper focuses on unary GEMM designs for integer-based DL inference and performs a detailed evaluation of three latest unary design proposals, namely, uGEMM, tuGEMM and tubGEMM, by comparing them to a conventional binary GEMM. Rigorous post-synthesis evaluations beyond prior works are performed across varying bit-widths and matrix sizes to assess the designs' tradeoffs and determine optimal sweetspots. Further, we perform weight sparsity analysis across eight pretrained convolutional neural networks (CNNs) and the LLaMA2 large language model (LLM). In this work, we demonstrate how unary GEMM can be effectively used for energy-efficient compute in future edge AI accelerators.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑