Author: Halyna Hyryavets

Confidential computing is a very topical issue that is gaining in importance with the growing popularity of cloud services, and this aspect is also important in the provision of HPC (High-Performance Computing) services. These tools and techniques help providers ensure the security of client data during their processing.

Competitors' role as "service providers" was to design and train an appropriate deep neural network (DNN) and / or gradient-boosted decision tree (GBDT) model on the data set provided. The condition was that any end user - the client - could use this model to process their data without the provider being able to decode / decrypt it.

More information about the challenge that hackers could work on can be found here.

During the hackathon, participants also had the opportunity to participate in a workshop on high-performance computing, where one of our lecturers explained how high-performance computing works, what are the areas of use and what are the possibilities of access to HPC infrastructure in Slovakia.

The team from the Spacecode Technical University in Košice won the award for the solution design - congratulations! 

Aftermovie

• 

• 

• 

• 

• 

• 

eINFRA and experience from the Czech Republic

Experts from CESNETwere invited to share their experiences with both NCCs in this area. CESNET is an association of universities of the Czech Republic and the Czech Academy of Sciences. It operates and develops the national e-infrastructure for science, research and education which encompasses a computer network, computational grids, data storage and collaborative environment. During the workshop, representatives of the CESNET presented such topics as:

M. Ruda: Introduction, CESNET and Metacentrum: large-scale distributed computing infrastructure and services.

D. Antoš: CESNET and Metacentrum: operation and development of data storage as a part of the national infrastructure for science and research.

Cloud services as an (un)necessary part of the HPC provider portfolio – is it profitable?

Branislav Jansík from IT4Innovations and the head of the Czech competence center Tomáš Karásek presented the requirements of users for access to cloud services. It is the motivation, requirements and expectations of users, as well as the benefits and barriers from the point of view of the HPC service provider, have been the subject of a longer discussion. Participants also debated alternative platforms, "easy-access" services, microservice, HPC "market".

Both National Competence Centers plan to integrate HPC cloud services into their portfolios using the open-source platforms OpenNebula and OpenStack. CESNET representatives who have experience with both tools shared their benefits and technical support options.

The more general debate concerned in particular the issue of systematic and long-term support for the development and operation of HPC systems as a necessary component of the national research infrastructure.

• 

• 

• 

• 

• 

• 

• 

• 

• 

The program begins with a several-day introductory online training, where participants learn more about HPC and get to know the entire team. In the coming weeks, they are working on individual projects, the results of which will be summarized in a short report and at the same time shown in the form of a video or presentation during the final mini-conference. The organizers also appreciate if students share their experiences on the Summer of HPC blog or through social media during the program. In the end, the best projects are awarded in the Best Visualization or Best Project categories.

Registration is open until April 12, 2022. All interesting project topics prepared by HPC experts this year can be found HERE.

Participants who will eventually participate in the program will be selected by a committee, which will also allocate a project based on their interest. One participant can choose 3 projects he would like to work on. Applicants can be from different disciplines, no previous experience with HPC is required - all you need is a desire to learn new things in the field of supercomputing!

PRACE Summer of HPC

The workshop will take place in person in the manor house Mojmírovce in Slovakia and participants will have the opportunity to gain a comprehensive theoretical overview and practical experience with quantum Monte Carlo methods and their molecular (Quantum Package and CHAMP codes) and periodic systems (TurboRVB). Lectures and trainings will be led by experts who are part of the European Center of Excellence TREX:

• Ivan Štich, Institute of Physics SAS
• Anthony Scemama, CNRS
• Claudia Filippi, Universiteit Twente
• Michele Casula, CNRS
• Sandro Sorella, SISSA

Main goals:

• Provide participants a general overview of QMC foundations and their adaptation to simulations of molecular and periodic systems.
• Provide an overview of the hot current topics in molecular, crystalline and low-dimensional systems. 
• Provide hands-on experience with practical QMC simulations based on CHAMP and TurboRVB simulation codes.
  

REGISTRATION

More information

The 9th annual cybersecurity community event Qubit Conference Prague 2022 offers its delegates excellent speakers, leading key cyber topics, keynotes, case studies, educational sessions, panel discussions and popular networking events. Two full days of top cybersecurity community event, providing both in-person and virtual experiences, are designed to give you a mix of opportunities for learning, networking, and collaboration.

Qubit Conference Prague is a cybersecurity community event connecting the East and West. Over the past years, Qubit Conference has grown to be a leading cybersecurity event in CEE region, bringing the latest cybersecurity news and knowledge to the community of cyber professionals. Following the previous eight years of success, the conference is expanding further and is bringing the conference back to Prague. Being held again in the “Rooftop of Europe” and Bohemian capital on May 25 – 26, community partners and cyber professionals from all over the world gather for two full days of engaging and inspiring educational sessions brought by some of the brightest minds in the cyber space.

Qubit Conference Prague 2022 is a unique way to meet the best and the brightest minds in the information security fields across multiple industries, and all carrier levels.

What you can expect:

• 220+ C-level managers, experts & practitioners at a unique cyber community event
• Excellent speakers and educational sessions
• Popular networking events, VIP receptions and exclusive meeting opportunities
• Pre-conference hands-on trainings
  

Qubit’s mission is to bring together and build a community of knowledge and information sharing for the industry’s experts and professionals through networking & education events. That is the reason why Qubit Conference aims to create space for the community of cybersecurity professionals to join in on inspiring, highly professional technical and strategical presentations and educational sessions, enjoy trainings with professional approach and experience friendly community atmosphere surrounded by key decision makers and experts in the field of cybersecurity.

Attending Qubit Conference Prague 2022 is an excellent opportunity to educate yourself, promote your company and meet new partners and fellow colleagues from the cybersec market. Connect with the event today and explore the latest innovations and ideas that are carving tomorrow‘s industry landscape.

Thanks to the media cooperation of the National Competence Center for HPC and Qubit Conference, you can now use the 10% discount code: QP22EUROCC10.

JOIN QUBIT CONFERENCE PRAGUE 2022!

The workshop will take place in person in Slovakia and participants will have the opportunity to gain a comprehensive theoretical overview and practical experience with quantum Monte Carlo methods and their molecular (Quantum Package and CHAMP codes) and periodic systems (TurboRVB). The lectures and trainings will be led by experts who are part of the European Center of Excellence TREX.

Main goals:

• Provide participants a general overview of QMC foundations and their adaptation to simulations of molecular and periodic systems.
• Provide an overview of the hot current topics in molecular, crystalline and low-dimensional systems. 
• Provide hands-on experience with practical QMC simulations based on CHAMP and TurboRVB simulation codes.
  

Registration will be open soon!

More information

Part of the lectures is planned as a sequence focusing on the applications of Machine Learning (ML) and Artificial Intelligence (AI). Come and listen to Slovak experts and take the opportunity to discuss with them in an informal atmosphere!

We would like to invite scientists, students and also the general public - all those who are interested in the topic of supercomputing. Research and development relying on high-performance calculations has multidisciplinary character and the knowledge of the basic modeling and simulation methods, programming languages and operating systems can be of a great advantage. This semester we will first look at the quantum Monte Carlo methods, considered very promising for utilizing the next generation exascale supercomputers. We will learn about the important role the high performance computers play in weather forecasting. Subsequently, we will focus on machine learning and artificial intelligence topic with lectures covering the basics as well as specific applications.

We plan to organize the lectures in a hybrid mode so it is possible to attend them in person. However, due to the unpredictable pandemic situation and constantly changing rules for mass events, we will inform you about the lecture format via our website and social networks approximately one week before each individual event.

You may look forward to the following lectures and speakers - and we look forward to see you!

• Modelling of 2D systems properties using the Quantum Monte Carlo methods – Ivan Štich z Fyzikálneho ústavu SAV (15. februára)
• Use of HPC in operational weather forecast in Slovakia - Jozef Vivoda from the Slovak Hydrometeorological Institute (March 1)
• Introduction to artificial neural networks and evolutionary algorithms – Ivan Sekaj z Ústavu robotiky a kybernetiky, FEI STU (15. marca)
• Deep learning and its basic applications for computer vision – Andrej Lúčny z Katedry informatiky FMFI UK (29. marca)
• Úvod do aplikácií zložitých výpočtov v medicíne – Zuzana Černeková z oddelenia počítačovej grafiky a videnia, Katedra aplikovanej informatiky FMFI UK a Peter Bluska z Rádiológia s.r.o.  (12. apríla)
• TheUse of ML / AI for applications in chemistry - potential drugs on Covid-19 – Marián Gall z Výpočtového strediska SAV (26. apríla)
• Current problems of reinforcement learning – Michal Chovanec zo spoločnosti  Tachyum (10. mája)
• Natural and artificial intelligence – Tomáš Hrodmádka z Neuroimunologickeho ústavu SAV (24. mája)
  

Registration

Facebook COO SAS

YouTube COO SAS

Účastníci získali prehľad o základných pojmoch a definíciách, ako operačný systém umožňuje a zabezpečuje prístup aplikácií a procesov k prostriedkom a ako tieto prostriedky spravuje. Dozvedeli sme sa viac o stavoch a zmenách stavu jednotlivých procesov. Zaujímavé boli aj informácie o tom, ako vyzerá politika a réžia plánovača (scheduler), ktorý prístup na CPU procesom prideľuje.

Z hľadiska vysokovýkonných výpočtových prostriedkov nás zaujíma, ako funguje OS v prostredí s mnohými procesormi. Zvyšovanie počtu procesorov je prirodzenou odpoveďou na rýchly nárast požiadaviek aplikácií a súčasne limitov zvyšovania výkonu jediného CPU. Operačný systém teda môže úlohy rozdeľovať medzi viaceré fyzické procesory (alebo jadrá), pričom tieto procesy sú nezávislé a môžu bežať súbežne. Programátori môžu využiť výhody viacerých procesorov a svoje úlohy rozdeliť na viacero súbežných podúloh. Tieto už ale nie sú nezávislé a väčšinou je potrebné, aby medzi sebou navzájom komunikovali. Jedna úloha – proces môže mať teda viacero samostatných tokov riadenia, ktoré nazývame vlákna (threads) a ktoré zdieľajú väčšinu prostriedkov tohto procesu, vrátane pamäte. Z pohľadu architektúry to môže vyzerať ako na obrázku ilustrujúcom schému symetrického multiprocesorového systému (Obr. 2), kde je jedna pamäť zdieľaná viacerými rovnocennými procesormi (architektúra UMA – Uniform Memory Access). Tu je potrebné ošetriť synchronizáciu prístupu k tejto spoločnej pamäti, čo je možné urobiť viacerými spôsobmi. Prístup SMP – symetrického multiprocesingu má však nevýhody ako zlá škálovateľnosť, čakanie pri synchronizácii, tzv. cache trashing.

Ak fyzickú pamäť rozdelíme na viaceré moduly, dosiahneme menšie zaťaženie zbernice, pretože procesory budú najviac využívať vlastnú lokálnu pamäť. Tento prístup poznáme ako NUMA – Non-Uniform Memory Access. Tu je najvýhodnejšie, ak OS alokuje úlohe pamäť pre dáta čo najbližšie k procesoru, na ktorom úloha beží. Na obrázku vidíme príklad prepojenia 4 procesorov:

Prednáška pokryla aj tému správy pamäte, virtuálnu pamäť a jej alokáciu – vrátane konceptu overcommit a tému súborového systému a jeho hierarchie.

Operačné systémy a ich fungovanie v HPC prostredí by si určite zaslúžili aj viac priestoru, ako náš formát môže poskytnúť. Ak vás téma zaujala, na Fakulte matematiky, fyziky a informatiky UK na túto tému prednáša práve Dr. Dušan Bernát – a ak ste našu prednášku nestihli, môžete si ju pozrieť na Facebook or YouTube.

Schedule and Registration

•  November 23, 17:00 - Short guide to software parallelization
  Jaroslav Suchánek (Biomedical Research Center, SAS)
•  December 7, 17:00 - Federated distributed computing
  Ladislav Hluchý (Institute of Informatics SAS)

More information about the series

• 

• 

The following panel discussion, organized with the support of the NCC, highlighted the situation in HPC in Slovakia. Lukáš Demovič, who is the director of the Centre of Operations,  Slovak Academy of Sciences - an institution operating the only publicly available HPC infrastructure – spoke about the current possibilities, ongoing projects and ambitious plans for significant expansion of computational capacities in Slovakia. The importance of investments and support of this field was stressed also by other speakers. Prof. Ivan Štich, who is active within the TREX CoEshowed use cases of 2D materials modelling using the QMC method,, which is, thanks to it’s robustness and scalability, especially suited for utilizing the coming exascale systems. Prof. Mária Bieliková from the Kempelen institute of intelligent technologies introduced a newly developed large neural language model for Slovak. She also mentioned the aspect of large energy consumption that comes with many AI applications and models together with the strategies scientists employ to mitigate this problem. Jozef Vivoda described models for numerical weather forecasting, employed and (co)developed at the Slovak hydrometeorological institutetheir level of precision as well as computational demands. Robert Cisár from the Tachyum company and Luboš Kolář from HPE took part in the discussion as representatives from the private sector.

All speakers agreed on the necessity of investing into HPC technologies at the national level and the importance of supporting the whole ecosystem at the same time. The lack of concept on the state level and the lengthy and disproportionately complicated process of public procurement were identified as main problems when planning and purchasing new supercomputers. The National competence centre is focusing it’s efforts primarily on building the supporting ecosystem and raising the adoption of HPC tools and connected technologies across all sectors. Without an available HPC infrastructure, however, this task remains a challenge.

Speakers' presentations:
How is Slovakia doing with supercomputing? – Lukáš Demovič.
HPC and numeric weather forecast in Slovakia -Jozef Vivoda.
Who can afford artificial intelligence today? – Mária Bieliková.
Is a supercomputer suitable only for science? – Robert Cisár.
2D semiconductor modeling by exascale computing methods – Ivan Štich.
The supercomputer is not just about computing power – Luboš Kolář.
Why do we need supercomputers? Discussion.

Computational power increased thanks to new technologies and innovations in semiconductor device fabrication and enabled doubling the number of components per integrated circuit every two years – this trend is described by the Moore’s law. However, we are now approaching physical limits in this area.

Another way of speeding up calculations is to use parallel approach, supported by today’s supercomputer architectures, including high-speed interconnecting network. Doc. Šperka described the history from the first parallel vector computers and Cray-1 supercomputer up to modern supercomputers.

Lecture series is a joint activity of the Computer museum COO SAS and National Competence Center for HPC. NCC, as the primary contact point for HPC in Slovakia, covers, among other things, educational and dissemination activities in this area.

You can watch the recording of the event on Facebook and YouTube and you can already register for our next lectures. Schedule and registration:

•  10. novembra 2021 o 17:00 – Operation systems in multiprocessor clusters
  Dušan Bernát (Faculty of mathematics, physics and informatics, Comenius University)
•  November 23, 17:00 - Short guide to software parallelization
  Jaroslav Suchánek (Biomedical Research Center, SAS)
•  December 7, 17:00 - Federated distributed computing
  Ladislav Hluchý (Institute of Informatics SAS)
•  December 21, 17:00 - Modelling of 2D systems properties using the Quantum Monte Carlo methods
  Ivan Štich (Institute of Physics SAS)

More information about the series

Registration

• 

• 

• 

• 

• 

•