CPU partition
The CPU partition consists of 92x HPE Cray EX425 Compute Blade, each blade holding 2 node. The CPU partition contains a total of 184 node.
Processor:
AMD EPYC 7763 64-Core Processor (2.45GHz)
Max. Boost Clock: Up to 3.5GHz
2x CPU per node
4x CPU per blade
Memory:
DDR4 3200MHz 16GB
8 DIMM module per socket
128 GB memory per socket 256 GB memory per node 512 GB memory per blade
Network:
HPE Slingshot 200GbE
CPU blade and node naming convention in the system:
CPU nodes are located in the x1000 cabinet.
For example: x1000c0s0b0n0
c - Chassis (1-8)
s - Slot (1-8)
b - Board (Blade) (0-1)
n - Node (0-1)
How to use CPU nodes
Example for an CPU interjactive job:
srun --partition=cpu --ntasks=1 --cpus-per-task=32 --mem-per-cpu=2000 bash your_job.sh
With this command, we run our your_job.sh job on 1 CPU node/32 CPU core, allocating 2000 MB memory for each CPU core.
How to run a CPU batch job:
The previous interactive job can also be run as a batch job. In this case, the content of the batch_script.sh will be the following:
#!/bin/bash
#SBATCH -A ACCOUNT
#SBATCH --partition=cpu
#SBATCH --job-name=jobname
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=32
#SBATCH --mem-per-cpu=2000
your_job.sh
This script can be queued with the following command:
sbatch batch_scipt.sh
According to our batch script, we will run our your_job.sh job on 1 CPU node/32 CPU core, allocating 2000 MB memory for each CPU core.
Further information about the hardware: