Tycho Supercomputer
Tycho is funded by a large collaboration of physicists at UCPH and tailor made to support computer driven scientific discovery including large-scale modelling, creating predictions for specific experiments and observations, and data intensive analysis.
Current areas supporting Tycho include astrophysics, biophysics, astro-particle physics, and gravity.
What was to become the Tycho supercomputer was created by Åke Nordlund in 2001 with funding from the Danish Center for Scientific Computing (DCSC), and the hardware was renewed over the next decade.
In 2012, DCSC was replaced by DeiC (Danish e-infrastructure Cooperation), and since then the hardware has been maintained through competitively obtained external funding with facility support by the faculty of SCIENCE.
In 2015, the size of Tycho and number of research groups started grow, and today more than 100 researchers primarily from the Niels Bohr Institute use Tycho to support their science.
Tycho is in the top five of the largest supercomputers in Denmark and the largest machine dedicated to a narrow group of researchers.
To access Tycho you must be a member of one of the research groups who have sponsored the hardware.
Please contact your group leader if you would like to know more. It is possible for new research groups to become a member of Tycho by contributing with hardware in reasonable proportion to their needs.
Please contact us if you are interested.
Tycho contains three types of computing infrastructure:
- Frontend machines are used to connect to the machine, prepare and develop software, analyze and investigate datasets, and do interactive work. We currently have four frontend machines with 512GB to 1024GB of memory, 32 to 64 CPU cores, very fast NVMe scratch disk space (up to 30 TB per machine), and powerful GPUs for GPU computing and visualization tasks.
- Compute nodes, spanning three different generations of CPU architectures, are used to carry out large scale computing. We have in total more than 10,000 compute cores with more than 50 TB of memory spread out over 140 machines. All machines are interconnected with fast infiniband network making it possible to carry out tightly coupled parallel computations using many compute nodes simultaneously for a single job.
- GPU nodes, with NVIDIA A100 GPUs and large amounts of memory for GPU computing.
The computational infrastructure is complemented with a 1,300 TB high performance storage system based on Lustre. Current peak performance is 443 TFlops from the CPUs and 93 TFlops from the GPUs.
We maintain a technical documentation site for Tycho here: https://wiki.nbi.ku.dk/tycho
Tycho Supercomputer
Head: Troels Haugbølle, Associate Professor
Phone: +45 29 38 25 88, Email: haugboel@nbi.ku.dk
Sponsors:
Mauricio Bustamente, Vitor Cardoso, Amin Doostmohammadi, Jose Ezquiaga, Troels Haugbølle, Anders Johansen, Uffe Gråe Jørgensen, Michiel Lambrechts, Åke Nordlund, Martin Pessah, Johan Samsing, Irene Tamborra
Contact:
Troels Haugbølle
Phone: +45 29 38 25 88
Email: haugboel@nbi.ku.dk