Spicecellar Compute Cluster
Last Update, Jusy 30, 2026
Welcome to Spicecellar, the group's private high-performance computing
cluster. It was built in-house to give the group fast, dedicated access
to compute for our simulation work — no queueing on shared university
resources, no waiting on external allocations.
What's inside
Spicecellar is six nodes in total: a login node, a storage server, and three dedicated compute nodes — Burgunder, Carmenere, and Riesling . The compute nodes are built for serious workloads: high core-count AMD EPYC processors, up to 597 GB of RAM on our largest node, and fast local NVMe storage. Riesling also carries an NVIDIA RTX A4000 GPU — GPU-accelerated jobs are being enabled and will follow.
What it's for
The group is rolling out a shared software stack for first-principles calcualtions,
micromagnetic and atomistic simulations, ML models — supporting the group's work on
computational materials design and antiferromagnetic and altermagnetic spintronics.
Live status & documentation
Current jobs, hardware utilization, and full documentation are tracked
outside this page, on the cluster's own site:
→ Go to the Spicecellar dashboard (spicecellar.de)
Access
If you're part of the group and need access to the cluster, contact
vebharad@uni-mainz.de, sinova@uni-mainz.de.