Discoverer Supercomputer: Unleashing High-Performance Computing for Bulgarian Tech

In the digital era, computational capacity is a primary driver of scientific innovation and economic competitiveness. From modeling global climate patterns and simulating molecular interactions for drug discovery to training complex deep learning architectures, modern research requires immense processing power. Recognizing this strategic imperative, the European Union established the European High-Performance Computing Joint Undertaking (EuroHPC JU) to construct a world-class supercomputing network across Europe.

A key hub in this European network is located in Sofia, Bulgaria. Housed inside Sofia Tech Park, the Discoverer Supercomputer is a petascale system that offers massive computational resources to academic researchers, technology startups, and international enterprises. Discoverer has established Bulgaria as the leading center for high-performance computing (HPC) in Southeast Europe.

Here is a comprehensive analysis of Discoverer’s technical architecture, access models, scientific and commercial use cases, and its strategic position within Europe’s supercomputing network in 2026.

Technical Architecture: Inside the BullSequana XH2000

The Discoverer Supercomputer was designed and built by Eviden (an Atos company), utilizing their advanced BullSequana XH2000 architecture. The system is engineered for maximum computational density, energy efficiency, and high-speed data transfer.

Core Technical Specifications

• Processing Power: Discoverer possesses a peak performance of over 4.5 Petaflops (4.5 quadrillion floating-point operations per second).
• Processors: The system is powered by AMD EPYC processors, featuring over 144,000 computing cores operating in parallel.
• System Memory (RAM): Equipped with over 300 Terabytes of high-speed RAM, enabling the system to process massive datasets in memory without storage bottlenecks.
• Storage Capacity: Features a high-performance Lustre storage system with over 2 Petabytes of capacity, optimized for high-speed read and write operations.
• Cooling Infrastructure: Discoverer utilizes an advanced direct liquid cooling system. Water-cooled cold plates contact the hot components directly, allowing the system to run at high efficiency with low energy consumption and a low Power Usage Effectiveness (PUE) ratio.

graph LR
A["Discoverer Supercomputer (4.5 Petaflops)"] --> B["Academic Research Track"]
A --> C["Enterprise R and D Track"]
A --> D["EuroHPC Infrastructure Synergy"]

B --> B1["Free Access via Open Science Peer Review"]
B --> B2["Biomedicine and Molecular Simulations"]

C --> C1["Paid Private Access with Full Confidentiality"]
C --> C2["Automotive CFD and Aerospace Simulations"]

D --> D1["BgGPT AI Model Training"]
D --> D2["Sofia Tech Park Ecosystem Integration"]

Access Model: Who Can Use Discoverer?

Discoverer is designed as a shared resource, accessible to both academic researchers and commercial enterprises. The access model is structured under EuroHPC guidelines to ensure fair allocation of computing resources:

The Academic and Scientific Track (Free Access)

Researchers from public universities and scientific institutes across Europe can apply for computing time on Discoverer free of charge.

• Application Process: Researchers submit detailed project proposals outlining their scientific goals and the computational resources required.
• Peer Review: Proposals are evaluated by an independent scientific committee based on their research quality, feasibility, and societal impact.
• Open Science Requirement: The results of the research must be published in open-access scientific journals, ensuring the knowledge benefits the wider scientific community.

The Commercial and Enterprise Track (Paid Access)

Private businesses, tech startups, and industrial manufacturers can purchase computing resources on Discoverer to run proprietary simulations, train machine learning models, or run big data analytics.

• Data Security: Commercial projects operate under strict confidentiality agreements. The data and code remain the exclusive property of the client, completely isolated from public access.
• Consulting Support: Discoverer’s local engineering team provides consulting support to corporate clients, helping them optimize their software code to run efficiently across parallel computing nodes.

High-Performance Computing (HPC) Systems in Europe

To understand Discoverer’s position in the European supercomputing landscape, consider this comparison of key EuroHPC systems:

While systems like LUMI and Leonardo offer massive pre-exascale capacity, Discoverer’s petascale architecture is highly optimized for regional industrial R&D, biotech modeling, and AI training.

Key Scientific and Commercial Use Cases

Discoverer is actively used for a wide range of advanced computing projects:

Biomedicine and Drug Discovery

Developing new medicines requires simulating how millions of chemical compounds interact with specific proteins in the human body. Running these molecular simulations on standard computers would take years. Discoverer allows researchers to run virtual molecular screening (in silico screening) in days, identifying promising drug candidates for cancers, viral diseases, and rare disorders, reducing drug development timelines.

Climate Modeling and Weather Forecasting

The National Institute of Meteorology and Hydrology uses Discoverer to run advanced climate models. By processing high-resolution atmospheric data in real-time, the supercomputer improves the accuracy of weather forecasts, tracks extreme weather events, and helps public authorities prepare for floods or droughts.

Automotive Simulation and Computational Fluid Dynamics (CFD)

Automotive component manufacturers in the Trakia Economic Zone partner with Discoverer to run CFD simulations. These simulations analyze how air flows around vehicle parts, optimizing aerodynamic efficiency, reducing wind resistance, and improving fuel efficiency for next-generation electric vehicles.

Training Large-Scale AI Models

Training models like BgGPT requires processing terabytes of text data across hundreds of GPUs or high-speed CPUs. Discoverer provides the computational pipelines required to train deep learning models, run neural network simulations, and process complex natural language data.

Joining the HPC Ecosystem: Sofia Tech Park Integration

Discoverer’s physical location in Sofia Tech Park integrates it with the wider technology ecosystem. Startups and R&D centers located in the park can connect directly to the supercomputing facility, working alongside research teams from Sofia University and the Bulgarian Academy of Sciences. This physical clustering facilitates the transition of academic research into commercial startup products, driving technology investments in Bulgaria.

Frequently Asked Questions: Discoverer Supercomputer

What is the processing performance of the Discoverer supercomputer?

Discoverer boasts a peak performance of over 4.5 Petaflops (4.5 quadrillion floating-point operations per second), powered by BullSequana XH2000 hardware with 144,000+ AMD EPYC cores, 300+ TB RAM, and 2+ PB high-speed Lustre storage.

Can private technology companies and startups rent computing power on Discoverer?

Yes. Under EuroHPC guidelines, private enterprises and startups can purchase compute hours on Discoverer for commercial R&D, AI model training, and fluid dynamics simulations under strict data confidentiality agreements.

How does Discoverer support AI model development in Bulgaria?

Discoverer provides the high-performance parallel computing nodes required to train large-scale artificial intelligence models, such as INSAIT’s BgGPT, as well as complex computer vision algorithms for autonomous systems.

What role does Discoverer play in the European EuroHPC network?

Discoverer is one of Europe’s primary petascale supercomputers, acting as a regional computational powerhouse for Southeast Europe, complementing pre-exascale systems like LUMI (Finland) and Leonardo (Italy).

Conclusion: Driving Regional Innovation

The Discoverer Supercomputer has transformed Bulgaria’s technological landscape. By providing over 4.5 Petaflops of computational power, the system enables academic researchers and private enterprises to run complex molecular simulations, climate models, and AI training pipelines locally. As a key hub in the EuroHPC network, Discoverer supports regional innovation, helps companies reduce R&D cycles, and positions Bulgaria at the forefront of high-performance computing in Southeast Europe.