一句话：通过线程韧性识别与重映射，将同韧性线程归入同一 warp，实现 GPGPU 选择性复制容错并降低开销。
GPU 广泛用于加速计算，但瞬态硬件故障（软错误）可能破坏输出。本文利用 GPGPU 应用中线程、warp 和协作线程阵列的层次组织，提出一种线程韧性识别与重映射方法：将具有相同韧性特征的线程映射到同一 warp，从而只在 warp 级别对真正需要保护的线程组启用部分复制机制，实现错误检测/纠正。作者在来自 4 个基准套件的 12 个基准（17 个内核）上评估，显示线程可被重映射为可靠或不可靠 warp，平均仅引入 1.63% 开销；线程重映射到不同 warp 不牺牲应用性能；该重映射支持 warp 复制，相比标准双副本和三副本平均分别减少 20.61% 和 27.15% 执行周期。
Graphics Processing Units (GPUs) are widely used by various applications in a broad variety of fields to accelerate their computation but remain susceptible to transient hardware faults (soft errors) that can easily compromise application output. By taking advantage of a general purpose GPU application hierarchical organization in threads, warps, and cooperative thread arrays, we propose a methodology that identifies the resilience of threads and aims to map threads with the same resilience characteristics to the same warp. This allows to engage partial replication mechanisms for error detection/correction at the warp level. By exploring 12 benchmarks (17 kernels) from 4 benchmark suites, we illustrate that threads can be remapped into reliable or unreliable warps with only 1.63% introduced overhead (on average), and then enable selective protection via replication to those groups of threads that truly need it. Furthermore, we show that thread remapping to different warps does not sacrifice application performance. We show how this remapping facilitates warp replication for error detection and/or correction and achieves average reduction of 20.61% and 27.15% execution cycles, respectively comparing to standard duplication/triplication.
DOI 原文 · arXiv · PDF（开放获取） · DBLP
@inproceedings{YangNJS21,
title = {Enabling Software Resilience in GPGPU Applications via Partial Thread Protection},
author = {Lishan Yang and Bin Nie and Adwait Jog and Evgenia Smirni},
booktitle = {ICSE},
year = {2021},
doi = {10.1109/icse43902.2021.00114}
}