arXiv cs.AI· Yousef Sadegheih, Dorit Merhof, Muhammad Usman·· 9 天前AI 评分22
DEX：面向高能效 MSDF 神经网络推理的数字级提前退出机制
DEX: Digit-Level Early Exit for Energy-Efficient MSDF Neural Network Inference
AI 导读
针对量化 U-Net 脑肿瘤分割，研究者提出 MSDF 加速器 DEX，通过数字级提前退出等四种运行时机制，在 BraTS 上把数字周期减少 38.38%，平均 Dice 达 80.58%（浮点模型为 81.20%）。
正文 · 原文
Abstract:U-Net inference for brain-tumor segmentation requires billions of multiply-accumulate operations, motivating hardware that can reduce computation dynamically rather than relying only on fixed precision or static model compression. Most-significant-digit-first (MSDF) arithmetic exposes the leading digits of a result during computation, enabling output-dependent decisions before the full value is generated. This paper presents an MSDF accelerator for quantized U-Net segmentation with a two-stage grouped processing element supporting signed INT8 operands and in-stream bias accumulation. Four runtime mechanisms operate directly on the output digit stream: exact early negative detection (END) in ReLU layers, exact sign-only decision making in the segmentation head, calibrated low-order-digit skipping, and calibrated pruning. The two approximate mechanisms are selected offline under an accuracy constraint, while execution requires only lightweight control and does not modify the stored weights. On a residual U-Net trained with nnU-Net for BraTS, the proposed mechanisms reduce digit cycles by 38.38\% while achieving a mean Dice score of 80.58\% on 73 held-out cases, compared with 81.20\% for the floating-point model; the exact mechanisms alone reduce cycles by 18.79\% without altering the quantized output. Synthesized in 45~nm, the processing element operates at 500~MHz, occupies 0.858~mm$^2$, and consumes 0.726~mJ per $192\times192$ patch under switching-activity-annotated power analysis. A projected eight-output accelerator with shared activation delivery achieves 16.6~ms latency and 1.67~mJ per patch.
| Subjects: | Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.11748 [cs.AR] |
| (or arXiv:2610.11748v1 [cs.AR] for this version) | |
| https://doi.org/10.48550/arXiv.2610.11748 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Muhammad Usman [view email]
[v1]
Thu, 8 Oct 2026 11:39:01 UTC (3,477 KB)
来源：arXiv cs.AI · arxiv.org