Karolina

Karolina featured image

System Overview

  • Name: Karolina
  • Location: Czech 
  • Owner: IT4Innovations at VSB – Technical University of Ostrava
  • Operator: IT4Innovations at VSB – Technical University of Ostrava
  • Released in: 2021
  • Primary focus area: Research

Architecture and Performance

Karolina consists of 829 computational nodes of which:

  • 720 are universal compute nodes HPE Apollo 2000 Gen10, 2 x AMD Zen2 EPYC 7H12, 2.6 GHz with 256 GB RAM
  • 72 are NVIDIA A100 accelerated nodes HPE Apollo 6500, 2 x AMD Zen3 EPYC 7763, 2.45 GHz with 1024 GB RAM and 8 x NVIDIA A100 (40 GB HBM2)
  • 1 is a data analytics node HPE SuperDome Flex system, 32 x Intel Xeon-SC 8268, 2.9 GHz with 24 TB RAM
  • 36 are cloud partitions HPE Apollo 2000 Gen10. 2 x AMD Zen2 EPYC 7H12, 2.6 GHz with 256 GB

User access to Karolina is provided by four login/access nodes.
The nodes are interlinked through high-speed InfiniBand and Ethernet networks.
Scratch storage is done by ClusterStor E1000 with 1PB with a throughput of 730.9 GB/s write and 1198.3 GB/s read.

Achievements and Impact

Rankings & records:
The petascale system Karolina, acquired as part of the EuroHPC Joint Undertaking, was installed in June 2021.
In the TOP500 list, which evaluates supercomputers in terms of their performance, it ranked 69th worldwide and 19th in Europe.
In the Green500 list of the most energy-efficient supercomputers, it ranked 8th in 2021.

Scientific contributions:
Pavel Hobza stands a great chance at rewriting the textbooks of physical chemistry
Researchers Enhanced Ammonia Production Process
Czech scientists become the first to observe an inhomogeneous electron charge distribution on an atom

Sustainability and Future Outlook

Environmental impact:

Karolina is running in an energy-efficient mode, which means:
Universal compute nodes’ CPUs run with a frequency capped at 2.1 GHz.
The accelerated partition’s GPUs’ streaming multiprocessor frequency is limited to 1.29 GHz. CPUs are not capped.
The expected impact on user codes’ performance and energy savings are provided to users in the IT4I documentation.

Future developments: upcoming hardware upgrades, integration of quantum computing, focus on AI/ML advancements etc.

Contact Information