Characterization-Guided GPU Fault Resilience in NVIDIA MPS

Rixin Liu 1 , Xingqi Cui 1 , Kaijian Wang 1 , Xinheng Ding 2 , Zirui Liu 2 , Yuke Wang 1 , Jiarong Xing 1 · 2026

Read Characterization-Guided GPU Fault Resilience in NVIDIA MPS by Rixin Liu 1 , Xingqi Cui 1 , Kaijian Wang 1 , Xinheng Ding 2 , Zirui Liu 2 , Yuke Wang 1 , Jiarong Xing 1 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-26461.txt.

Description

Rixin Liu and six colleagues harden NVIDIA Multi-Process Service, which shares one graphics processor across processes, against faults that kill every co-running process. A survey of graphics-processor faults guides the design. Memory faults get full isolation through software work in the open driver kernel module, keeping fail-stop behavior. Other faults fall back on a recovery base sharing state via virtual memory and pre-started standbys. Tests show no outage for isolation and full throughput back in 355 milliseconds after recovery.

Catalog record

CREATORRixin Liu 1 , Xingqi Cui 1 , Kaijian Wang 1 , Xinheng Ding 2 , Zirui Liu 2 , Yuke Wang 1 , Jiarong Xing 1
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE70913
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.26461

Browse the full library · About the Tower of Babel