Research processing at FGV now has high-performance computing (HPC) environment

This new environment was opened by FGV's Division of Operations, through its Information and Communication Technology Sector, and benefited FGV Energy, which managed to drastically reduce the time needed to simulate operation planning and the expansion of the electric sector in Brazil
机构
06 八月 2020
Research processing at FGV now has high-performance computing (HPC) environment

In July, Fundação Getulio Vargas’ Division of Operations (FGV DO), through its Information and Communication Technology (ICT) Sector, opened a high-performance computing (HPC) environment to conduct research. Because of its architecture, the HPC environment can deliver a greater computational capacity than would normally be delivered by conventional architecture.

FGV’s first center to benefit from this new environment was FGV Energy, which managed to drastically reduce the time needed to simulate operation planning and the expansion of the electric sector in Brazil. When it first processed CEPEL models on a virtual machine, FGV Energy’s simulation took more than 24 hours to complete. After the delivery of the HPC environment, the same simulation took around three hours, matching the time taken on market platforms.

Having obtained these good results in its first experience, FGV Energy plans to use the HPC platform for other projects. Given that the regulatory framework governing the country’s power sector is currently being changed, studies and simulations that can reflect and anticipate socioeconomic impacts are in ever-greater demand.

This platform was set up by FGV DO’s ITC Sector in a pioneering way, on the Microsoft Azure Cloud, allowing it to be used and priced on demand. Its main feature is its “elastic” environment, which can incorporate resources in line with research needs.

To begin with, the environment is enabled to run C++ and Fortran codes, and it will soon be able to use other languages, such as MATLAB, R and Python. Thanks to this innovation and the successful experience of its first test, FGV is able to support research that requires HPC, as well as demands from other processing platforms, such as Hadoop, Cosmos DB and Databricks.

Nosso website coleta informações do seu dispositivo e da sua navegação por meio de cookies para permitir funcionalidades como: melhorar o funcionamento técnico das páginas, mensurar a audiência do website e oferecer produtos e serviços relevantes por meio de anúncios personalizados. Para saber mais sobre as informações e cookies que coletamos, acesse a nossa Política de Cookies e a nossa Política de Privacidade.