The high-performance computer lab for big data analysis at the University of Évora (UÉ) invested in two Nvidia DGX A100 devices with a global capacity of 10 petaflops (10,000,000,000,000,000 instructions per second), 16 GPU A100, 640 GB of memory GPU and 2 TB of RAM. This system provides a performance equivalent to 10 thousand computers and is unique in Portugal, representing an investment of 400 thousand euros.

“We are endowing this laboratory with Artificial Intelligence and Big Data with high computational capacity in the field of automatic learning and artificial intelligence”, explains, in a statement, Paulo Quaresma, professor at the Informatics Department at the University of Évora and who heads the BigData Laboratory @HUH. The maximum total processing capacity of this equipment is 10 petaflops, equivalent to about ten thousand desktop computers without a specialized graphics card.

The Alentejo 2020 program finances this project and the equipment will be used to “strongly enhance the development of research, innovation and knowledge transfer in artificial intelligence and big data”. A competition is underway to hire a doctoral researcher for three years to ensure the correct management and enhancement of this equipment, UÉ also says.

One of the projects in which the university is participating, financed by the Foundation for Science and Technology (FCT), is the development of an AI-based system that will reduce the time taken to answer each phone call to the university by 5%. Line SNS24. The current application of the methodologies allows “to classify the calls in one of the 59 existing algorithms with a degree of correction greater than 80% and to identify the three most likely algorithms with a degree of accuracy of 97%”, explains Paulo Quarema. The acquisition of the new computer system will allow the application of methodologies based on more modern deep learning networks and there is an expectation that it will be possible to further improve the performance already achieved.

BigData @ UE is a high performance computational laboratory for analyzing large data sets and which aims to provide research and innovation services not only in the Alentejo region, but also at national and international level.

Leave a Reply

Your email address will not be published. Required fields are marked *