Exploring AMD GPU scheduling details by experimenting with “worst practices”

Exploring AMD GPU scheduling details by experimenting with “worst practices”
复制标题

通过试验“最差实践”来探索 AMD GPU 调度细节

DOI:
10.1007/s11241-022-09381-y
复制
发表时间:
2022
期刊:
影响因子:
1.3
通讯作者:
Anderson, James H.
Anderson, James H.
中科院分区:
计算机科学3区
文献类型:
--
作者:
Otterness, Nathan;Anderson, James H.

文献摘要

参考文献

被引文献

相似文献

图形处理单元(GPU)是最近实时研究的重要目标,但研究往往受到GPU硬件和软件的“黑匣子”性质的阻碍。现在,一个GPU制造商,AMD,已经接受了一个开源软件栈,人们可能会期望增加的实时研究使用AMD GPU。然而,现实更为复杂。在不了解内部细节可能存在差异的情况下,研究人员无法假设使用NVIDIA GPU进行的观察将继续适用于AMD GPU。此外,AMD软件的开放性并不意味着他们的调度行为是显而易见的,特别是由于稀疏,分散的文档。在本文中,我们将不同的文档集合到一个连贯的源中,提供了如何在AMD GPU上调度计算工作的端到端描述。在此过程中,我们首先具体演示了不正确的管理如何在共享AMD GPU中触发极端最坏情况的行为。随后,我们将解释AMD GPU的内部调度规则,它们如何导致“最差实践”,以及如何正确管理AMD GPU共享中一些最关键的性能因素。
Graphics processing units (GPUs) have been the target of a significant body of recent real-time research, but research is often hampered by the “black box” nature of GPU hardware and software. Now that one GPU manufacturer, AMD, has embraced an open-source software stack, one may expect an increased amount of real-time research to use AMD GPUs. Reality, however, is more complicated. Without understanding where internal details may differ, researchers have no basis for assuming that observations made using NVIDIA GPUs will continue to hold for AMD GPUs. Additionally, the openness of AMD’s software does not mean that their scheduling behavior is obvious, especially due to sparse, scattered documentation. In this paper, we gather the disparate pieces of documentation into a single coherent source that provides an end-to-end description of how compute work is scheduled on AMD GPUs. In doing so, we start with a concrete demonstration of how incorrect management triggers extreme worst-case behavior in shared AMD GPUs. Subsequently, we explain the internal scheduling rules for AMD GPUs, how they led to the “worst practices,” and how to correctly manage some of the most performance-critical factors in AMD GPU sharing.
DOI: 10.1109/rtss.2018.00021
发表时间: 2018-12
期刊: 2018 IEEE Real-Time Systems Symposium (RTSS)
影响因子: --
作者:
Nicola Capodieci;R. Cavicchioli;M. Bertogna;Aingara Paramakuru
通讯作者: Nicola Capodieci;R. Cavicchioli;M. Bertogna;Aingara Paramakuru
AMD GPU 作为 NVIDIA 的替代品来支持实时工作负载
DOI: 10.4230/lipics.ecrts.2020.10
发表时间: 2020
期刊: Proceedings of the 32nd Euromicro Conference on Real-Time Systems
影响因子: --
作者:
Otterness, N.;Anderson, J.
通讯作者: Anderson, J.
DOI: 10.1109/rtss.2018.00018
发表时间: 2018-12
期刊: 2018 IEEE Real-Time Systems Symposium (RTSS)
影响因子: --
作者:
Ming Yang;Tanya Amert;Kecheng Yang;Nathan Otterness;James H. Anderson;F. D. Smith;Shige Wang
通讯作者: Ming Yang;Tanya Amert;Kecheng Yang;Nathan Otterness;James H. Anderson;F. D. Smith;Shige Wang
剖析 CUDA 调度层次结构:性能和可预测性视角
DOI: 10.1109/rtas48715.2020.000-5
发表时间: 2020
期刊: 2020 IEEE Real-Time and Embedded Technology and Applications Symposium (RTAS)
影响因子: --
作者:
Ignacio Sañudo Olmedo;Nicola Capodieci;Jorge Martinez;A. Marongiu;M. Bertogna
通讯作者: M. Bertogna
NVIDIA GPU 上的逆向工程电源管理 - 自主就绪系统剖析
DOI: --
发表时间: 2013
期刊:
影响因子: --
作者:
M. Peres
通讯作者: M. Peres