一句话：首次对超大规模数据中心向量指令静默数据损坏进行指令与应用级分析，并提出低开销检测机制。
静默数据损坏（SDC）会绕过检测并传播错误结果，Google、Meta、阿里巴巴等生产系统已有报告。本文首次对超大规模数据中心中向量指令的 SDC 进行指令级与应用级综合分析。作者采用两阶段方法，在超过 140 亿 CPU 秒内执行逾 78 万亿测试轮，发现此前未记录的 SDC 模式，为理解潜在成因和设计缓解策略提供依据。基于这些观察，论文提出一种低开销 SDC 检测机制，利用应用内基于算法的容错。评估表明，该方法即使对中等规模输入也能实现 88% 到 100% 的 SDC 机器检测率，时间开销仅 1.35%。
Silent Data Corruption (SDC) poses a reliability threat in modern datacenters. These insidious errors evade detections and propagate incorrect results throughout the system. Companies including Google, Meta, and Alibaba have reported SDC incidents affecting their production. In this paper, we present the first comprehensive instruction- and application-level analysis of vector instruction SDCs in hyper-scale datacenters using a two-stage approach. We perform over 78 trillion test rounds in more than 14 billion CPU seconds. Our observations reveal undocumented SDC patterns that provide insights into possible underlying causes and inspire new mitigation strategies. Based on these findings, we propose a low-overhead SDC detection mechanism leveraging in-application algorithm-based fault tolerance. Our method achieves 88% to 100% SDC machine detection rate with a time overhead of only 1.35% even for modestly sized inputs.
@inproceedings{MeiVDSR26,
title = {SEVI: Silent Data Corruption of Vector Instructions in Hyper-Scale Datacenters},
author = {Yixuan Mei and Shreya Varshini and Harish Dixit and Sriram Sankar and K. V. Rashmi},
booktitle = {ASPLOS},
year = {2026},
doi = {10.1145/3779212.3790217}
}