A country that does not have the computational capacity to solve its own problems will have no alternative but to try to acquire solutions provided by others. One of the most important aspects of sovereignty in the 21st century is the ability to produce mathematical models and to have the capacity to solve them. Today, the availability of computing power commensurate with one’s wealth exponentially increases a country’s capacity to produce knowledge. in the developed world, it is estimated that for every dollar invested in supercomputing, the return to society is of the order of US$ 44(1) and to the academic world US$ 30(2). For these reasons, HPC occupies an important place on the political and diplomatic agendas of developed countries. 

In Latin America, investment in HPC is very low compared to what’s the US, Asia and Europe are doing. In order to quantify this difference, we present the tables below, which show the accumulated computing capacity in the ranking of the 500 most powerful supercomputers in the world – the TOP500(3) – (Table 1), and the local reality (Table 2). Other data are also included, such as the population (in millions), the number of researchers per 1,000 inhabitants (Res/1000), the computing capacity per researcher (Gflops/Res) and the computing capacity per US$ million of GPD. In Table 1, we have grouped the countries by geographical area. America appears as the area with the highest computing capacity, essentially due to the USA, which has almost 45% of the world’s computing capacity in the TOP500. It if followed by Asia and then Europe. Tis TOP500 list includes mainly academic research centres, but also industry ones, typically those used in applied research (many private ones do not wish to publish such information for obvious reasons). For example, in Brazil – which shows good computing capacity with 88,175 TFlops – the vast majority is in the hands of the oil industry and only about 3,000 TFlops are used for basic research. Countries listed in the TOP500 invest in HPC from a few TFlops per million GDP (Belgium 5, Spain 7, Bulgaria 8), through countries investing in the order of hundreds (Italy 176, Japan 151, USA 138), to even thousands, as is the case in Finland with 1,478. For those countries where we were able to find data on the number of researchers, these range from a few Gflops per researcher (Belgium 19, Spain 24, Hungary 52) to close to 1,000 GFlops, i.e. 1 TFlop (USA 970, Italy 966), with Finland surpassing this barrier with 4,647. Note that, unlike what happens locally, countries with a certain degree of development invest every 3-4 years in supercomputing, so the data we are showing will soon be updated and there will be variations in the list. For example, this year a new supercomputer will come into operation in Spain(4), which, with an investment of some 150 million euros, will give Spain one of the most powerful supercomputers in Europe – and the world.

Table 1. HPC availability per researcher and relative to GDP in the TOP500 countries (includes HPC in industry).

The local reality is far from this data. Table 2 shows data from Argentina, Brazil, Chile and Mexico. In Chile, the availability of computing power is 2-3 times less per researcher than in countries with less computing power in the OECD and up to 100 times less than a researcher in the US. In Chile, our investment measured in TFlops per million US$ of GDP is 166 times less than in the US; with respect to European countries that invest less in HPC it is 9 times less, and with respect to the European average (including Finland) it is 80 times less, i.e. the difference is considerable. It is clear that we need to close this gap. An investment go about 5 million dollars in HPC infrastructure in the next 5 years would close this gap by a factor of almost 20 times our computational capacity. However, returning to the example of Spain, the supercomputer it will have this year will offer 23 times more computing power than at present and, therefore, we will only maintain our relative distance. If we do not invest, the dap will increase by at least 23 times and will end up being huge. Therefore, we do not only need a one-time investment, but we need to ensure a regular investment. Some neighbouring countries are already investing significantly in supercomputing. This is the case in Argentina, where they are investing 7 million dollars (2 million for the datacenter and 5 million to buy a new supercomputer), which will increase their current capacities by almost 40 times(5).

Table 2. HPC availability per researcher and relative to GDP in the region (*only HPC capacity in academia is considered in this table).

For the above reasons, we are working to convince the Chilean authorities that we must have greater funding and, more crucially, permanent state funding in HPC. In relation to this, on July 6 we signed a collaboration agreement between 44 institutions with the support of the Ministry of Science to work on the creation of the National Supercomputing Laboratory(6). The agreement recognised that supercomputers are a critical infrastructure for Chile’s development, that it is necessary to centralise the requirements/resources at the national level, obtain permanent funding from the State and create a new institutional framework to provide governance. In an unprecedented inter-institutional collaboration in Chile, the competition for HPC resources at the national level is eliminated ad the possibility of direct funding from the State is opened up without generating controversy.

Undoubtedly, supercomputing is a fundamental pillar for the development of any country, where increasing investment provides a strategic advantage, and in Latin America we should not be left behind.

By NLHPC

References

(1) Hyperion Research HPC Investments Bring High Returns

(2) EESI-2 Special Study To Measure And Model How Investments In HPC Can Create Financial ROI And Scientific Innovation In Europe 

(3) https://top500.org/ 

(4) https://www.lavanguardia.com/ciencia/20230129/8713515/llega-superordenador-marenostrum-5-bsc-barcelona.html

(5) https://www.hpcwire.com/2022/12/15/argentina-announces-new-supercomputer-for-national-science/

(6) https://uchile.cl/noticias/187955/44-instituciones-crearan-el-laboratorio-nacional-de-supercomputacion

The post Towards a greater HPC capacity in Latin America first appeared on RISC2 Project.

The NLHPC staff  set their  fundamental pillar to focus  efforts on the scientific community and show HPC as an essential tool for country development by getting users from diverging scientific areas, industry and public sector. This entails breaking access barriers to this kind of technology. NLHPC faces this challenge by making training for the basic use of HPC  and scientific software optimization;  which is key in order to make a good use of resources.

The training was carried out within a framework of computational thinking, being the process by which an individual, through his professional experience and acquired knowledge, manages to face problems of different kinds. This could be evidenced in our active participation in the resolution of the proposed activities, which enhanced our abstraction and engineering thinking. We will certainly take this vision of education and collaborative work to our professional environment, in the different roles we play as HPC administrators, teachers and students.

The proper use of computing services involves efforts to perform monitoring, control and infrastructure management tasks. With the help of the tools reviewed during our visit, we will be able to provide our users with the highest standards of quality, security and accessibility.

The joint effort of the RISC2 and EU-CELAC ResInfra projects made it possible for engineers from Colombia, Mexico and Peru to participate in this HPC management course, learn about Chilean culture, gain knowledge and valuable contacts for our profession.

After living this great experience, we hope that in the near future other supercomputing centers replicate this type of initiatives in other parts of the world, thus increasing the communication bridges between HPC administrators from different places, sharing knowledge and experiences.

We are left with the milestone of being part of the First School of HPC Administrators of Latin America and the Caribbean, with experiences that made us grow in professional, academic, and human aspects. As well as with alliances among colleagues and now friends, a network of support as brothers of the same region.

We conclude by thanking Rafael Mayo of CIEMAT for the initiative; Ginés Guerrero, Pedro Schürmann, Eugenio Guerra, Pablo Flores, Angelo Guajardo, Esteban Osorio, José Morales for the knowledge and experiences shared; RISC2 and EU-CELAC ResInfra for providing us with this learning opportunity, supporting the scholarship grant.

By:

Miguel Angel Barrera Arbelaez, Universidad de los Andes, Colombia

Carlos Enrique Mosquera Trujillo, Centro de bioinformática y biología computacional de Colombia BIOS, Colombia

César Alexander Bernal Díaz, Universidad Industrial de Santander, Colombia.

Eduardo Romero Arzate, Universidad Autónoma Metropolitana, México.

Ronald Darwin Apaza Veliz, Universidad Nacional de San Agustín, Perú.

Joel Gonzalez Lara, Centro de Análisis de Datos y Supercómputo, México

The post First School of HPC Administrators in Latin America and the Caribbean: A space for the formation of computational thinking first appeared on RISC2 Project.

This school aimed to train HPC sysadmins with the latest technologies in supercomputing in a two-week training program, and discussed different topics, such as compilations, visualization and monitoring tools, networking, security tools, and installation, configuration and use of SLURM and EasyBuild, among many others.

According to Ginés Guerrero, the Executive Director of the NLHPC and one of the organizers of this training, “the NLHPC team wanted to pass on the knowledge gained for more than a decade to other administrators, so they can benefit from our experience. This has involved a great effort by a team of 7 engineers, putting aside all their tasks for several weeks to prepare an intensive 64-hour school from scratch. In addition, this process has been tailor-made, since the students indicated their own interests through a form.”

In total, the event had 8 participants from various countries: 2 from Mexico, 3 from Colombia, 2 from Chile, and 1 from Peru, leveraging the international networking opportunities and promoting closer relations between the administrators of various supercomputing centers in Latin America, the main goal of the RISC2 project. A team of 7 engineers (Guinés Guerrero, Pedro Schürmann, Eugenio Guerra, Pablo Flores, Ángelo Guajardo, Esteban Osorio, and José Morales) from NLHPC was responsible for delivering all the 35 lectures.

The post RISC2 supported the first school of HPC Administrators in Latin America and Caribe first appeared on RISC2 Project.

The post First School of HPC Administrators in Latin America and Caribe first appeared on RISC2 Project.

The post GUACOLDA – LEFTRARU first appeared on RISC2 Project.

The fund, granted by the Chilean National Agency for Research and Development, provides the possibility to finance and maintain the center’s team of engineers for the next five years, with the chance of renewal. If everything is going well and according to plan, the Center will be granted funding for another five years. The fund gives NLHPC the confidence to continue to grow and reach their objectives.

“Collaboration with international networks was important in the evaluation of this call. Undoubtedly, the fact of being part of RISC2 has been very relevant for us to be awarded the project”, stated Ginés Guerrero, NLHPC Director.

The NLHPC is the national supercomputing center in Chile, with specialization in high performance computing, that manages Guacolda-Leftraru, the most powerful supercomputer in South America. This infrastructure appeared in November in the IO500 list, a ranking of the fastest supercomputing (HPC) storage systems in the world, where it was placed in 39th place.

During the last year, more than 400 users from different areas of knowledge, more than 45 institutions have used the NLHPC. The center is at the service of the national scientific community, industry, and State and its main goal is to satisfy the demand for high performance computing.

Photo: Comunicaciones FCFM- U. de Chile

The post The National Laboratory for High Performance Computing awarded funding for 5 years first appeared on RISC2 Project.

This funding will help continue the work of the NLHPC, while strengthening its supercomputing infrastructures, which have been crucial to the national scientific community.

It is important to mention that the submission to the contest was supported by the RISC2 project, since the NLHPC’s team received support letters from the RISC2 consortium, including BSC, RedCLARA, RICAP, and SCALAC.

According to ANID, “the II Contest for Major Scientific and Technological Equipment Fondequip seeks to install scientific capacity, covering the country’s need for a larger and more extensive infrastructure, which houses sophisticated and innovative scientific equipment, promoting and facilitating the development of research excellence in the national territory with this scientific and technological resources. It also seeks to position Chile at the forefront of research excellence and frontier research at the international level”.

About NLHPC

The NLHPC specializes in HPC and manages Guacolda-Leftraru, the most powerful supercomputer in Chile and one of the most powerful in South America.

The NLHPC is at the service of the national scientific community, the State, and the industry that requires HPC services. Its main mission is to meet the national scientific demand for high performance computing, providing high-quality services and promoting its use in both basic and applied research problems.

The center began to take shape in 2009 as an initiative of the Center for Mathematical Modeling (CMM) at the Faculty of Physical and Mathematical Sciences of the University of Chile, which invited different institutions to join efforts towards the creation of a national supercomputing laboratory. The response of these institutions was very enthusiastic, and their support allowed the center to start operating in 2011. Since then, most research institutions in the country are part of the center, constituting the largest scientific network in Chile sharing an infrastructure. The laboratory has also signed collaboration agreements with several state institutions, international networks, and supercomputing centers around the world.

Photo: Guacolda-Leftraru, by National Laboratory for High Performance Computing (NLHPC)

The post National Laboratory for High Performance Computing awarded Fondequip Major funding first appeared on RISC2 Project.

CARLA is an international conference aimed at providing a forum to foster the growth and strength of the High-Performance Computing (HPC) community in Latin America through the exchange and dissemination of new ideas, techniques, and research in HPC and its applications areas. The general chair of the 2021 edition is Isidoro Gitler, from Cinvestav, who coordinates and participates in all the event’s activities.

Workshops

Different partners of the RISC2 were involved in the organization of the scientific workshops as part of the CARLA 2021 conference. Between seven workshops organized to this conference, two of them are from the RISC2 consortium. The workshop on HPC Collaboration between Europe and Latin America took place online, on October 5, 2021. The goal was to provide a space dedicated to exchange experiences, towards the promotion and support of new collaborations across different countries of Europe and Latin America, within the framework of the recently launched ‘A network for supporting the coordination of High-Performance Computing research between Europe and Latin America’ (RISC2). This workshop achieved a maximum number of 55 participants, with Pedro Vieira Alberto, from the University of Coimbra, as one of the Invited Speakers. Ulisses Cortés, from Barcelona Supercomputing Center, presented the RISC2 project, as an example of HPC collaboration between Europe and Latin America. The chairs of this event were Ulisses Cortés and Rafael Mayo-García, from CIEMAT.

Another workshop organized by the RISC2 team was the workshop on HPC an Energy, which was held online on October 4, with Álvaro Coutinho, from COPPE, as Chair. This workshop focused on HPC techniques applied to the energy sector, in order to improve and reform many industrial sectors. HPC can provide several solutions to the energy sector, e.g., oil and gas solutions in upstream, midstream, and downstream problems; improve wind energy performance; solve issues of combustion efficiency for transportation systems; making nuclear systems more efficient and safer; improving solar energy systems; optimizing wind energy systems; improving the quality and efficiency of seismic and geophysical simulations, etc.

It is important to mention that Ginés Guerrero, from NLHPC, was one of the chairs of workshops organized by CARLA 2021.

Tutorials

Carla Osthoff, from Laboratório Nacional de Computação Científica, was Tutorial Chair Member of the 11 tutorials that were accepted at the CARLA 2021. CARLA 2021 provided tutorials and hands-on workshops for both introductory and advanced levels, specifically designed to undergraduate and master students all over Latin- American countries. There were two different periods: Fundamental Tutorials, which included six different tutorials the week before CARLA 2021, and Advanced Tutorials, with five different tutorials held a week after CARLA 2021. CARLA 2021 Tutorials were supported by Latin American, Caribbean, and European institutions.
Esteban Mosckos, from the Universidad de Buenos Aires, was involved in the organization of two different tutorials: “OpenMP: Introduction to shared memory models” and “Introduction to Distributed Memory Models using MPI”. Both activities consisted of theory and hands-on exercises, lasting four hours, with close to 40 assistants each.

The NLHPC partner was also responsible for a tutorial focused on Working with a resource manager on an HPC infrastructure, for the use of SLURM. Two more tutorials were organized by NLHPC, including, the tutorial on Performance Analysis Tools, with the participation of one of the members of the NLHPC Scientific Committee, and the other on Quantum Computing, with the participation of IBM.

The CARLA 2021 conference had more than 30 institutions on the board committee and more than 115 in attendance, connected simultaneously.

All the videos are available here.

The post RISC2 with a strong presence at CARLA 2021 first appeared on RISC2 Project.