NCC BULGARIA

The National Competence Centre of Bulgaria (NCC-Bulgaria) in the area of High-Performance Computing (HPC), High-Performance Data Analytics (HPDA) and Artificial Intelligence (AI) has the goal to enhance and develop the competences of the Bulgarian computational community, making full use of EuroHPC resources and the EuroCC partnership.

The NCC-Bulgaria is built by a consortium coordinated by the Institute of Information and Communication Technologies at the Bulgarian Academy of Sciences (IICT-BAS), and two members, Sofia University “St. Kliment Ohridski” (SU), and University of National and World Economy (UNWE). The three partners carry diverse technical and scientific background in the area of HPC and ICT in general, so as to ensure achievement of the project objectives and guarantee the overall success. The partners collaborate with Sofia Tech Park, where the Discoverer EuroHPC supercomputer is operating.

Technical/scientific Challenge:

Collaborative research project funded by the Bulgarian Science Fund and performed by teams from the Institute of Information and Communication Technologies and the Institute of Neurobiology at the Bulgarian Academy of Sciences led to the development of a hierarchical model of the human brain visual system. It was implemented using NEST Simulator – an open source platform running on desktop, HPC, or supercomputer architectures allowing parallel simulation of user-defined structures of spiking neural networks models of any size. The running time of such a model on a desktop PC increased dramatically with its dimension. Besides, the overall model included also a Python module whose running speed appeared to be the bottleneck in simulation investigations. Although parallel implementation on a desktop PC with an 8-core processor using mpi4py allowed a significant drop in its run time, it still remained the slowest part of the model. Another challenge to be solved was related to merging of NEST module with mpi4py parallel simulation. Hence the need for more powerful computer architecture appeared obvious for further development and refinement of the overall model.

Solution:

Since supercomputers offer much more powerful computing hardware than single processor desktop PCs, implementation of the model on supercomputer Avitohol appeared the proper solution to the challenge described above. Three algorithms were developed and tested:

  • Algorithm 1: Python module parallelized into multiple processes.
  • Algorithm 2: Python module parallelized via spawning within a single process.
  • Algorithm 3: The combined run of NEST and Python modules in a parallel simulation using spawning.

Scientific impact:

Running of the parallel Python module on the HPC facility of IICT – the supercomputer Avitohol – allowed usage of a much higher number of cores and thus a higher number of parallel processes, so a significant drop in its computation time was achieved. The problem arising from merging NEST and Python modules in a common MPI environment was overcome using a dynamic process management version of the Python module (Algorithm 3). Thus the overall model run time was significantly decreased allowing for its further application for in-silico investigation of the human visual system and brain structures involved in it by varying their parameters.

Benefits: 

  • We developed a pipeline for HPC implementation of spike timing neural network models of various brain structures.
  • HPC implementation of such models will allow for a variety of simulation investigations of brain functioning.

Success story # Highlights:

  • Keywords: Supercomputer applications; hydrodynamic; High-dimensional simulations, Brain modelling
  • Research area: Neuroscience
  • Technology: HPC/AI

Speedup and efficiency of algorithms

This project has received funding from the European High-Performance Computing Joint Undertaking (JU) under grant agreement No 951732. The JU receives support from the European Union’s Horizon 2020 research and innovation programme and Germany, Bulgaria, Austria, Croatia, Cyprus, the Czech Republic, Denmark, Estonia, Finland, Greece, Hungary, Ireland, Italy, Lithuania, Latvia, Poland, Portugal, Romania, Slovenia, Spain, Sweden, the United Kingdom, France, the Netherlands, Belgium, Luxembourg, Slovakia, Norway, Switzerland, Turkey, Republic of North Macedonia, Iceland, Montenegro

Contact:

Prof. Petia Koprinkova-Hristova,
Assoc. Prof. Sofiya Ivanovska,
Assist. Prof. Simona Nedelcheva,
Mariya Durchova,
Institute of Information and Communication Technologies,
Bulgarian Academy of Sciences

Comparison of simulation times for different number of processes N and nodes