OMNI Cluster Usage Policy

Institute GPU Cluster(2 × NVIDIA B200, 2x H200, 1x A100 DGX)

IIIT Delhi | Version 2.0 | June 2026

OMNI Cluster Job Submission Guide

Queue Information

Settings Queue Type
Short Medium Long CAI
Default Time Limit 2 hr 6 hr 1 day 1 day
Maximum Time Limit 6 hr 1 day 3 days 3 days
Max Job Submit per Group 3 2 1 1
Storage quota 5 TB per Group
Max. CPU Cores 10 16 20 32
No. of available GPUs per job 1 1 1 2
Max. available RAM per job 64 GB 256 GB 512 GB 512 GB
Token Cost 0.1 per job 0.5 per job 1 per job 0.5 per job
Queue Description
Short MIG-enabled (8x 2= 16 GPUs @ 70 GB each) GPUs (DGX H200)
Medium Dedicated GPUs (B200 =8 GPUs @ 180GB , hgxh200=8 GPUs @144 GB )
Long Dedicated GPUs (B2002= 8 GPUs @192 GB)
CAI Dedicated GPUs (dgxA100= 8 GPUs @40 GB) - Currently Down

1. Short Queue

srun -A GroupName -p short -q short --ntasks=1 --gres=gpu:3g.71gb:1 --mem=64G --pty nvidia-smi

2. Medium Queue

srun -A GroupName -p medium -q medium --ntasks=1 --gres=gpu:1 --nodelist=hgxh200 --mem=60G --pty nvidia-smi

srun -A GroupName -p medium -q medium --ntasks=1 --gres=gpu:1 --nodelist=B2001 --mem=60G --pty nvidia-smi

3. Long Queue

srun -A GroupName -p long -q long --gres=gpu:1 --nodelist=B2002 --mem=60G --pty nvidia-smi

4. CAI Queue

Only valid for CAI Faculty Member Groups.

srun -A GroupName -p cai -q cai --gres=gpu:1 --nodelist=dgxA100 --mem=60G --pty bash

Job script

#!/bin/bash
#SBATCH --time=06:00:00
#SBATCH --job-name=Any_name
#SBATCH --account=GroupName
#SBATCH --partition=short
#SBATCH --qos=short
#SBATCH --ntasks=1
#SBATCH --gres=gpu:3g.71gb:1
#SBATCH --mem=64G

nvidia-smi > nvidialog.txt

Save the script as jobscript.slurm and submit using:

sbatch jobscript.slurm

Transparency

GPU allocation, utilization and ownership are visible to all users. Only resource usage information is public. Your source code and data remain private.

Prohibited Activities

  • Bypassing the GPU quota system.
  • Obtaining unauthorized administrative privileges.
  • Cryptocurrency mining.
  • Commercial or non-academic workloads.
  • Keeping GPUs idle intentionally.
  • Sharing login credentials.
  • Illegal or prohibited activities.

Monitoring and Enforcement

GPU jobs are logged, monitored and audited. Misuse may result in reduced quota, temporary suspension or permanent loss of access.

Support and Exceptions

  • Additional GPU hours may be requested with Faculty approval.
  • User tiers are managed by administrators.
  • Contact: helpdesk@iiitd.ac.in
Agreement

By using the GPU Server you agree to comply with this policy.