Petrobras runs biggest supercomputer in Latin America

Petrobras’s supercomputer Fênix is among the world’s 500 biggest computers and ranks first in Latin America. The list was compiled by Top500.org, based on the machines’ performance in data processing, and features Fênix at the 142nd position worldwide.

The computer features 55,296 gigabyte memory and 48,384 processing cores in its CPU. The machine, built by French manufacturer Bull, doubles the oil giant’s capacity for geophysical data processing. Petrobras General Manager Jonilton Pessoa said appearing on the list is important as it shows that the state-controlled company is still technologically up-to-date.

Fênix on the global stage

“Top500’s [list] gives us an idea of our standing vis-à-vis the rest of the world. In Latin America, we’re number one. This means we are a company with the appropriate processing capacity for its size,” Pessoa pointed out.

The company’s technological complex is always being brought up to date, he said. “We were innovative in the 00’s when we used GPUs, the graphics processing units, which brought us among the leaders in geophysical data processing back in 2012. Now all companies are using GPUs,” he remarked.

Even though computers were state-of-the-art back then, Pessoa noted, they are now outdated after seven years.

“During the crisis we faced, we were stagnant as far as the acquisition of new machines is concerned. Computer capacity kept making progress, and whatever was outstanding in 2012 does not have the same efficiency or efficacy nowadays. Some later updates did take place, but they were not on a par with our processing needs,” he explained.

Stepping up the game

Jonilton Pessoa said the tests that showed the time needed for processing a set of data can drop from eight to four months. The first results on a new production scale have not been released yet, as Fênix did not become operational until March. With the new applied algorithm, gains in speed can rise fourfold.

Geophysical data processing consists of transforming the information obtained in seismic acquisition—which comes originally as vibrations—into images. The technology identifies materials underground through the exchange of seismic waves in both land and sea.

This enables experts to locate salt layers, rock types, as well as oil, gas, and water reserves. Materials are detected through the properties with which waves return, like signal speed and amplitude.

Higher quality

The supercomputer, Pessoa went on to explain, can process data at a higher speed and expand algorithms that make the data more accurate, thus enhancing images—making oil reserves easier to spot and ascertaining where drilling can be best performed.

“This makes it possible for you to get a much better view of what’s below the surface. It looks better, so we can make a better decision when there’s an auction and base our decision and estimates on more accurate data,” he said.

The staff

In charge of operating Fênix are professionals from a number of fields, including IT workers, supervisors and analysts, some of whom work to translate mathematical formulas into a language to be read by the machine.

In geophysical technology, algorithms for data processing are developed, while geophysicists work to feed the computer with data from the seismic acquisition. Next, a team of geologists and geophysicists interprets the coming data and build a geological model for the area.

Another three on the way

Pessoa says the technological upgrade is part of the company’s plan to optimize its computer complex, under which three other machines the same model as Fênix were ordered.

“Our goal is to have 15 times as much processing capacity in 2020 than we did in 2018. We got sidetracked, and now we’ve been working to get back on track and never steer off again,” he concluded.