Slurm超额订购GPU [英] Slurm oversubscribe GPUs

问题描述

是否有一种方法可以在Slurm上超额订购GPU，即运行共享一个GPU的多个作业/作业步骤?我们仅找到了超额预订CPU和内存的方法，而没有找到GPU.

Is there a way to oversubscribe GPUs on Slurm, i.e. run multiple jobs/job steps that share one GPU? We've only found ways to oversubscribe CPUs and memory, but not GPUs.

我们要在同一个GPU上并行运行多个作业步骤，并可以选择指定用于每个步骤的GPU内存.

We want to run multiple job steps on the same GPU in parallel and optionally specify the GPU memory used for each step.

推荐答案

最简单的方法是将GPU定义为feature而不是gres，这样Slurm就不会管理GPU.确保需要一个工作的工作在提供一个工作的节点上.

The easiest way of doing that is to have the GPU defined as a feature rather than as a gres so Slurm will not manage the GPUs, just make sure that job that need one land on nodes that offer one.

这篇关于Slurm超额订购GPU的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！