Artelys Crystal HPC
A managed, sovereign high-performance computing platform for industry
Organisations running complex simulations, large-scale optimisations or compute-intensive workloads on sensitive data face a threefold challenge: securing the computing power their ambitions demand, guaranteeing data sovereignty, and avoiding weeks of configuration before obtaining their first results.
Artelys Crystal HPC addresses these three requirements head-on.
- Computing power is secured by dedicated, high-end infrastructure hosted at OVHcloud, whose fine-grained resource allocation makes it possible to pool software licences across teams rather than tying them up workstation by workstation.
- Data sovereignty relies on the option of hosting on SecNumCloud-qualified OVHcloud clusters located in France.
- Go-live is measured in hours, thanks to ready-to-use scientific environments and Artelys’ support on sizing and scaling your business-critical computations.
This is rarely the case with traditional HPC solutions, which impose trade-offs on security, flexibility or cost — trade-offs that slow teams down and expose organisations to risk. On-premise is rigid and expensive to scale; hyperscalers bring flexibility and power but raise sovereignty concerns; academic infrastructures are not built for recurring industrial use.
In practical terms, Artelys Crystal HPC is a fully managed HPC-as-a-Service platform, set apart by its user-centric approach: a single point of access, designed to move intensive computations into production quickly without sacrificing either performance or control over data.
A managed, sovereign high-performance computing platform for industry
Organisations running complex simulations, large-scale optimisations or compute-intensive workloads on sensitive data face a threefold challenge: securing the computing power their ambitions demand, guaranteeing data sovereignty, and avoiding weeks of configuration before obtaining their first results.
Artelys Crystal HPC addresses these three requirements head-on.
- Computing power is secured by dedicated, high-end infrastructure hosted at OVHcloud, whose fine-grained resource allocation makes it possible to pool software licences across teams rather than tying them up workstation by workstation.
- Data sovereignty relies on the option of hosting on SecNumCloud-qualified OVHcloud clusters located in France.
- Go-live is measured in hours, thanks to ready-to-use scientific environments and Artelys’ support on sizing and scaling your business-critical computations.
This is rarely the case with traditional HPC solutions, which impose trade-offs on security, flexibility or cost — trade-offs that slow teams down and expose organisations to risk. On-premise is rigid and expensive to scale; hyperscalers bring flexibility and power but raise sovereignty concerns; academic infrastructures are not built for recurring industrial use.
In practical terms, Artelys Crystal HPC is a fully managed HPC-as-a-Service platform, set apart by its user-centric approach: a single point of access, designed to move intensive computations into production quickly without sacrificing either performance or control over data.
The 5 core pillars
AGILITY
A platform at the service of your teams
PERFORMANCE
State-of-the-art computing power, without compromise
SECURITY
A sovereign SecNumCloud-ready platform
OBSERVABILITY
Full control,
in real time
OPTIMISATION
Licence pooling without compromising isolation
Inside the platform
AGILITY
Artelys Crystal HPC handles two complementary types of usage on a single platform: interactive sessions, where users explore, adjust and test their model in real time (via a remote desktop or a Jupyter Lab notebook, for example); and production jobs, compute-intensive runs launched in the background on full datasets. The platform’s orchestrator lets both coexist without friction, from exploration through to large-scale production.
-
- Collaborative environment : teams work together on a unified platform without interfering with one another’s processes.
- Seamless orchestration : run your simulations in parallel without worrying about resource conflicts or workflow collisions.
PERFORMANCE
In partnership with OVHcloud, Artelys Crystal HPC runs on high-end Bare Metal servers (up to 384 threads and 3 TB of RAM per node). The absence of a virtualisation layer eliminates performance overhead and, above all, allows fine-grained machine tuning through advanced settings — CPU frequency, turbo mode, optimised RAM configuration — ensuring optimal throughput for computation codes.
-
- End-to-end managed service : maintenance, obsolescence and hardware procurement are handled by Artelys — your teams don’t have to think about them.
- Seamless hardware upgrades : processor refreshes and configuration adjustments take place with no action and no reinvestment on your part.
- Proven scalability : the platform has demonstrated its ability to run industrial workloads comparable to those of the most advanced research infrastructures.
SÉCURITY AND DATA SOVEREIGNTY
Artelys Crystal HPC is designed to process intensive, sensitive workloads, with localised data management and strict access protocols.
-
- Data and workflow containment : for maximum protection, data and computations remain strictly confined to the cluster on which the platform is deployed. Nothing leaves the environment during processing.
- Secure access management : the platform uses single sign-on (SSO) for user authentication, ensuring centralised and tightly controlled access to the infrastructure.
- Deployment on SecNumCloud-qualified infrastructure : for the most sensitive or critical workloads, Artelys Crystal HPC can be deployed on OVHcloud Bare Metal Pod clusters qualified SecNumCloud by ANSSI, the French national cybersecurity agency.
OBSERVABILITY AND OPERATIONAL MONITORING
Artelys Crystal HPC natively integrates monitoring tools to give users full visibility over their compute workloads and resource consumption.
-
- Real-time tracking : Prometheus collects system metrics (CPU, RAM, disk I/O, network), surfaced through Grafana dashboards, allowing users to monitor job execution and the state of cluster resources.
- Secure monitoring infrastructure : access to the monitoring tools is strictly protected. Prometheus and Grafana sit behind a reverse proxy, with OAuth authentication for Grafana, ensuring that data is only accessible to authorised users.
- End-to-end visibility : every user retains direct access to their own data, logs and results across the entire lifecycle of a study.
COST OPTIMISATION
Artelys Crystal HPC reduces Total Cost of Ownership (TCO) by acting on two key financial components: competitive infrastructure pricing and pooled management of software licences.
-
- Competitive cloud pricing : a long-standing partnership with OVHcloud gives access to high-performance bare metal resources on favourable terms.
- Licence pooling : several teams share and centralise their software licences instead of duplicating them, delivering significant savings.
- Fine-grained resource control : SLURM makes it possible to reserve precisely the number of cores each job requires — a direct lever for optimising per-CPU licence costs, by avoiding the mobilisation of unused resources and therefore unused licences.
Our Use Cases
Concrete results for our clients :
Artelys Crystal HPC was used by Artelys as part of a study carried out with Artelys Crystal Super Grid, its energy system simulation software, on behalf of Litgrid for an adequacy assessment:
Compute time reduced from 48 hours to just 10 hours for more than 10,000 simultaneous Monte Carlo simulations.
Artelys Crystal HPC was used by Artelys as part of a study carried out with Artelys Crystal Super Grid, its energy system simulation software, on behalf of Elering for a security-of-supply assessment:
Orchestration of thousands of simulations to assess security-of-supply indicators as part of a cost-benefit analysis (CBA) for a critical cross-border electricity transmission line.
Artelys Crystal HPC was used by Artelys as part of a study carried out with Artelys Crystal Super Grid, its energy system simulation software, on behalf of AST for a security-of-supply assessment:
Orchestration of thousands of simulations to assess security-of-supply indicators as part of a cost-benefit analysis (CBA) for a critical cross-border electricity transmission line.
Support from Artelys’ HPC experts
Beyond the platform itself, Artelys experts support your teams on:
Sizing your infrastructure
Parallelising your code
Optimising your computing environments
Ready to upgrade your IT skills ?
contact us
© ARTELYS • All rights reserved • Legal mentions