A high-end fault-tolerant server designed for computation-intensive tasks. Thanks to its hybrid architecture based on CPU AMD EPYC and GPU NVIDIA A100, it is possible to achieve a performance density of 624 teraflop/s per 2U (for tasks using BFLOAT16/FP16 numeric formats).
Scope of application:
- Big Data Analysis – generation of analytical reports, processing of unordered collections of large data sets
- Mathematic modeling and simulation in science, medicine, and engineering design – DNA decoding, structure analysis of substances, materials, and medicines
- High-precision forecasting – weather forecasting, evaluation, and search of new mineral deposits
- Development of AI applications – AI programs and robotics
- Cryptography – data confidentiality and integrity
- Statistical data processing – evaluation and grouping of observations, calculation of the average values and indices
- Complex graphic objects – animation rendering, creation of visual effects, and motion picture production
- The server solution is easy to install and easy to use thanks to the logical arrangement of components and power units. Scaling is possible both by installing more data center servers and using more powerful and higher-capacity components.