HLRS test access

hlrs.de · Compute · 1 record · JSON

HLRS test access Compute

fails: identityinstitutionno cardDwrite-gated

www.hlrs.de/apply-for-computing-time · national HPC allocation · compute/hlrs-test-access · JSON · compare

Free tier	A test account on HLRS systems carries a contingent of 100 Hunter node hours for up to six months, granted on a one to three page application that is explicitly not peer-reviewed; GCS Preparatory Access gives a larger contingent plus HLRS user-support assistance for up to one year. unit hours; metered on wall-clock; exploratory / test allocation
Price	none stated
Per-object cap	not applicable
Retention	not applicable
Egress	not stated
Identity	institution not published
Limits	write-gated Test access is offered to potential applicants for GCS Large-Scale and rolling regular projects, which are German national calls for investigators at German research institutions, so the account presumes an eligible German home organisation. source Test access is offered to potential applicants for GCS Large-Scale and rolling regular projects, which are German national calls for investigators at German research institutions, so the account presumes an eligible German home organisation.
Binding	write-gated Test access is offered to potential applicants for GCS Large-Scale and rolling regular projects, which are German national calls for investigators at German research institutions, so the account presumes an eligible German home organisation.
Interface	unknown inferred from the tooling the provider documents; not probed
Tooling	not stated
Composes with	The un-peer-reviewed test account is the cheapest entry in the German stack, and it exists specifically to generate the scaling data a GCS regular or large-scale proposal needs.
Reliability	no reliability event recorded
Notes	HLRS is the only centre found that states outright that the test application is not peer-reviewed, which makes 100 Hunter node hours the most accessible German allocation; the eligibility section heading still ties it to German research institutions. Hunter carries AMD MI300A APUs, so 100 node hours is a small but real GPU allocation. GCS large-scale thresholds are useful as the corpus's clearest definition of what 'large' means in core hours versus GPU hours on the same programme.
Evidence rung	documented
Checked	2026-10-22 (live); stale on 2026-11-21

Evidence (3)

date	source	what it proved
2026-10-22	www.hlrs.de/apply-for-computing-time	'Within your test account you can consume a resource contingent of 100 Hunter node hours on the granted hardware platforms. The project time is limited to six months'; 'This short proposal will not be peer-reviewed', quoted
2026-10-22	www.hlrs.de/apply-for-computing-time	page section heading 'For investigators at German research institutions'; GCS Preparatory Access 'may last a maximum of one year' and provides 'a higher resource contingent than ordinary GCS test projects', quoted
2026-10-22	www.gauss-centre.eu/for-users/hpc-access/allocation-pgms	GCS large-scale threshold 'at least 25,000 node-h on Hunter', '5,250 EFLOP on JUWELS Cluster', '140,000 GPU hours on SuperMUC-NG Phase 2', equal to 2% of each system's annual production, quoted

More fields (1)

{
 "wave2": {
  "class": "national-hpc",
  "free_tier_kind": "exploratory / test allocation",
  "shape": {
   "vcpu": null,
   "ram_gib": null,
   "gpu_class": "AMD MI300A (Hunter), NVIDIA A100 (Hawk)",
   "gpu_count": null
  }
 }
}

Unknown or withheld values (10)

$.egress.metered egress-unknown
$.egress.terms not-stated
$.identity_cost.detail process
$.interface interface-unknown
$.per_object_cap.notes process
$.per_object_cap.stated not-applicable
$.reliability_note none-observed
$.retention.notes process
$.retention.stated not-applicable
$.tooling not-stated