A supercomputer is a term generally referring to a machine which ranks among the TOP 500 fastest high-performance computing systems worldwide. Such machines are created and owned by state as well as private research institutions, universities, and also companies.

It can be said that every scientific field will reach the point where conventional computational resources are not enough anymore and the computation would take too long. Experiments in laboratories are too expensive, and no scientist can, in terms of money and time, afford such a number of repetitions a supercomputer can simulate. What can then be computed by supercomputers? And what computations will our graduates be able to compute using these supercomputers?

It is focused on these problematic areas:

- Data analysis and data procession,
- Effective exploitation of HPC architectures,
- machine learning,
- Bioinformatics.

Numerical simulations and experiments in physics and chemistry have become an equal alternative to the real experiments in the last two decades in the development of e.g.

- new materials,
- pharmaceuticals, etc.

Mathematical modelling applications are realized on the supercomputers of IT4Innovations.

The typical problems are e.g.:

- development and implementation of parallel scalable algorithms of numerical mathematics.

Development and application of modern numerical methods in areas such as:

- aerodynamics and acoustics,
- shape optimization of components,
- crash test simulations etc.

Salomon supercomputer has been bought in the first half of this year. It has nicely ranked 40^{th} in July´s TOP 500. Salomon comprises 1008 computational nodes in total, 576 of which are common computational nodes and 432 accelerated computational nodes, each of which contains two Intel® Xeon Phi™ coprocessors. This fact makes Salomon the biggest installation of Intel® Xeon Phi™ coprocessors in Europe.

Detailed technical specification

Our small Anselm cluster was delivered to us by Bull company. Its theoretical power is 94 TFLOPS. Anselm, the computational cluster, consists of 207 computational nodes, 180 of which are standard computational nodes, 23 of which are accelerated by Kepler K20 GPU cards, and four of which are accelerated by the newest Intel® Xeon Phi™ technology. Moreover, it contains two thick nodes with higher operational memory capacity.

Detailed technical specification

Permoník is a mini-supercomputer comprising 8 Raspberry Pi cards, which can be fitted in a handbag. Despite its size, it demonstrates a real supercomputer, such as Salomon or Anselm, in terms of design, and it allows parallel computations via Ethernet connection and MPI library. Raspberry Pi card is in the size of a credit card, but its CPU has 700 MHz and its operational memory is 512 MB. As a cluster, Permoník has the power of 1GFLOPS. It is mainly used for promotion of supercomputing at events for students.

© Copyright 2015 by IT4I