AMD EPYC™ servers for brain dynamics research

AMD EPYC™ servers support brain dynamics research at Charles University

27. 8. 2024

The research team dealing with computational neuroscience at the Faculty of Mathematics and Physics of Charles University uses a server solution supplied by our company within the “Brain Dynamics” project. Newly deployed servers with AMD EPYC™ processors enable the development of computational models of brain activity that map brain dynamics in detail.

Developing computational models of brain activity

As part of the research work of the scientific group dealing with computational neuroscience, led by Mgr. Ján Antolík, Ph.D., the Department of Software and Teaching of Computer Science at the Faculty of Mathematics and Physics of Charles University uses a server solution supplied by our company. The research project “Brain Dynamics”, for which the servers are specifically used, is focused on the development of computational models of brain activity that capture the dynamics of the brain at different time scales. The several-fold increase in computational resource capacity will thus allow, within the framework of a specific research goal, a large-scale biophysically detailed evaluation of a neural network model.

Research project “Brain Dynamics”

This project is co-financed by the European Regional Development Fund (ERDF) through the Operational Programme Jan Amos Komenský. It is interdisciplinary in nature, combining neuropsychological experiments, advanced neuroimaging methods and non-invasive stimulation with computational models of brain dynamics and machine learning. Goals include developing flexible computational models for specific applications, monitoring and predicting brain conditions such as epilepsy, and optimizing medical brain stimulation and personalized drug therapy.

More information about the project

Supplied solution

The team of experts is now assisted by servers for the compute models that feature AMD EPYC™ 9384X processors with large L3 Cache, using AMD 3D V-Cache™ technology. Thanks to them, the entire cluster achieves great performance results – e.g. the SPEC2017FP rate reaches 4,810 points. Each server is built in a 1U Asus-branded design, codenamed RS700A-E12-RS12U, and includes two redundant power supplies. The cluster consists of 5 servers and the maximum power consumption is expected to be around 5 kW.

Parametr	Cluster
Počet CPU	10 × AMD Epyc™ Genoa-X 9384X
Celková velikost L3 Cache	7 680 MB
Použité CPU	AMD EPYC™ 9384X, 32 jader, základní frekvence 3,1 GHz (boost až 3,9 Ghz), L3 cache 768 MB, TDP 320 W
Celkový výkon	Výkon v pohyblivé řádové čárce dle testu SPEC2017FP rate 4 810 bodů
Celkový počet jader	320
Síťové připojení	Ethernet Dual Port 25GbE
Max. spotřeba	~ 5 kW
Provedení	1U

“Simulations by colleagues in the computational neuroscience group typically run as CPU-independent tasks (1 instance-per-socket). The computations are very memory bandwidth intensive and are greatly accelerated by the large L3 cache. To support 12 channels of DDR5 memory and a 768 MB “3D V-Cache”, we purchased 5 machines with a pair of 32-core AMD EPYC 9384X (24 MB cache per core) and 1.5 TB or 768 GB of RAM (12-channel or 6-channel with plans to purchase the remaining modules).”

– Mgr. Martin Mirbauer, Department of Software and Computer Science Education, Faculty of Mathematics and Physics, Charles University –

AMD EPYC™ with AMD 3D V-Cache™ technology (“Genoa-X”)

These CPUs are part of AMD’s EPYC™ processor family, which AMD has launched for servers and data centers. This particular model is part of the EPYC™ 9004 series, codenamed “Genoa-X” and is part of the fifth generation of AMD EPYC™ server processors based on the Zen 4 architecture. A key feature of the processors is AMD’s 3D V-Cache technology, which supports more than 1GB of L3 cache on a 96-core CPU. The larger cache enables faster CPU access to data needed for large and demanding computing tasks.

More information about processors

M Computers is AMD’s Elite Partner for EPYC™ processors.

AMD Partner

Check out our other case studies where AMD EPYC™ was part of the bundled solution and other articles dedicated to these processors.

More information

Testing

Are you interested in how AMD EPYC™ processors can be used in your infrastructure or would you like to experience the performance and features of these processors?
Please fill out the form or contact Lukáš Vach.

Lukáš Vach

Intel, AMD servers

M: 731 678 146 lukas.vach@mcomputers.cz