From mathematics and informatics, to medicine, chemistry, mechanical engineering, smart farming and agriculture, the Hemus supercomputer will serve many fields of knowledge. Together with Prof. Dr. Emanouil Atanassov from the Center for Excellence in Informatics and Information and Communication Technologies at BAS and Todor Todorov, Manager of Key Customers at Hewlett Packard Enterprise operated by Selectium, we discussed what applications this machine has and how the Bulgarian scientists and the Bulgarian society in general .can benefit from them. The new supercomputer has a performance of more than over 3 petaflops, provided by 148 servers, distributed in two subsystems based on HPE ProLiant servers. One subsystem uses universal multi-core processors embedded in 128 dual-processor HPE ProLiant XL220n Gen10+ servers with direct liquid cooling (Intel Xeon 8352Y). The second subsystem is GPU-based, with 20 dual-processor ProLiant XL675d servers using both general-purpose processors (AMD EPYC 7742) and Nvidia A100 graphics accelerators with Tensor cores, both types of chips being liquid-cooled. Nvidia’s graphics cards are particularly useful for developing fast algorithms for generating quasi-random sequences, which have numerous applications, including better image processing, climate change research, road traffic data, and more.

Hear more in the episode: