---
title: "Comment by Abhinav Sharma on \"KAI Scheduler Deep Dive: How NVIDIA Schedules GPUs on Kubernetes\" | Hashnode"
description: "The one-pod-at-a-time model is exactly why distributed training jobs stall on a busy cluster: part of the gang gets placed, the rest pend, and the placed pods h"
url: "https://hashnode.com/posts/kai-scheduler-deep-dive-how-nvidia-schedules-gpus-on-kubernetes/6ad9850f035d0b8ec4de8b8f/comment/6ad9a6d0124f36abb02219ae"
---

[Discussion](https://hashnode.com/posts/kai-scheduler-deep-dive-how-nvidia-schedules-gpus-on-kubernetes/6ad9850f035d0b8ec4de8b8f)

Comment [KAI Scheduler Deep Dive: How NVIDIA Schedules GPUs on Kubernetes](https://hashnode.com/posts/kai-scheduler-deep-dive-how-nvidia-schedules-gpus-on-kubernetes/6ad9850f035d0b8ec4de8b8f)

[Abhinav Sharma](https://hashnode.com/@abhinavs1920)

[Oct 22](https://hashnode.com/posts/kai-scheduler-deep-dive-how-nvidia-schedules-gpus-on-kubernetes/6ad9850f035d0b8ec4de8b8f/comment/6ad9a6d0124f36abb02219ae)

The one-pod-at-a-time model is exactly why distributed training jobs stall on a busy cluster: part of the gang gets placed, the rest pend, and the placed pods hold GPUs idle. Treating the whole job as one scheduling unit fixes the utilization problem, not just the placement one.
