Skip to content

Billing Policy

Allocations are measured in core-hours. One allocated CPU core for one hour is one CPU core-hour. GPU time is converted to equivalent core-hours using the offering's rules. Check your project usage for the reported balance.

The examples below describe reserved capacity. They do not establish billing weights or a final charge.

Good Practice

Example 1: A Balanced Half of a GH200 Node

One GPU, 72 CPU cores, and half the host memory form a proportional half-node request. Use less CPU or memory if your application needs less.

# Resource directives to use in your batch script:
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=72
#SBATCH --gres=gpu:1
#SBATCH --mem=480000M

The memory estimate uses 960,000 MiB per node; check scontrol show node for the current value. One hour reserves 72 CPU core-hours and 1 GPU-hour, before applying billing rules.

Example 2: Both GPUs and the Entire GH200 Node

For an application needing the full node:

# Resource directives to use in your batch script:
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=144
#SBATCH --gres=gpu:2
#SBATCH --exclusive
#SBATCH --mem=0

--exclusive reserves the node; --mem=0 requests all configured host memory. The application must support both GPUs and multiple CPU threads. One hour occupies 144 CPU core-hours and 2 GPU-hours of capacity.

Use these directives with a complete job template. Request both GPUs without exclusive access if you need less CPU or host memory.

Common Mistakes

Request Problem
--mem=0 for a small job Reserves all host memory and may prevent sharing
Too little CPU or host memory for a GPU Can stall data loading or cause failure
--exclusive by default Reserves capacity a small job may not use
More resources than the application supports Does not make a serial application parallel

Measure usage and adjust requests with a reasonable margin. --mem requests host memory, not GPU memory. Fewer CPUs do not remove GPU charges.

Follow the shared project limits. See Slurm batch options for allocation semantics.