“I am concerned that large companies like Google, Microsoft, Facebook or Amazon have almost infinite resources” | Technology | The USA Print

0


Jack Dongarra’s plans (72 years old next July) were to become a high school teacher in his native Illinois (United States). In his final year of college, however, computers got in his way: He received a fellowship at one of the nation’s leading science and engineering research laboratories, Argone National Laboratory. “That’s when I fell in love with them,” he says humble and smiling in a video conference from the basement of his home in the state of Tennessee. Half a century later, this scientist’s programming work has earned him the latest Turing Award for “pioneering concepts and methods that have given rise to computational calculations that have changed the world.” This award is endowed with one million dollars (financed by Google) and is considered by many to be something like the Nobel Prize for Computer Science. “It is a very great honor that I did not expect. My head is still spinning,” he admits.

Meteorological prediction, astrophysics, study of diseases, climate change, simulation of traffic accidents, wars… The power of supercomputers either high performance computers has exploded in the last decade, and there are few areas where they do not exert a decisive influence. “Supercomputing touches all citizens: today, it is computers that drive science,” says Dongarra. “We simulate things, we develop models and then these allow us to predict what is going to happen.” allow these super gadgets“precision machines”, in his opinion, “at the level of telescopes such as Hubble or James Webb”.

The moore’s law —Intel co-founder Gordon Moore’s 1965 assertion that the number of transistors on a chip doubles every two years—explains the explosion of computing in the last 40 years. But this exponential power would have been useless if there had not been an operating system on which to work. And that is where the figure of the recent Turing award and his contributions to algorithms and libraries emerge -the latter constitute the basis of modern programming: they function as a kind of software of support for other programs and without them, programmers would have to be rewriting all programs continuously—which allowed these languages ​​to keep pace with the exponential improvements of the hardware.

Image of the Fugaku supercomputer taken in 2020.
Image of the Fugaku supercomputer taken in 2020.STR (AFP)

academic baggage

Dongarra’s academic and school background is appreciated when he tries to explain the strength of a device with these characteristics. “It’s hard to imagine how fast these machines are,” he says. Currently, the most powerful on the planet is called Fugaku, it is owned by Fujitsu and is housed in Tokyo. Its power amounts to 415,000 teraflops (TFLOPS). The FLOPS—floating point operations per second—is the magnitude used to portray the extremely high performance of these devices. The Fugaku, therefore, is capable of 415,000 trillion operations per second.

“The Japanese computer is incredibly fast,” he admits. And yet, his reign may have little left. What is expected to become the fastest computer in the world is being built at the Oak Ridge National Laboratory, just over 80 kilometers from where Dongarra resides. It will be the first device in the world capable of exceeding exaFLOP (one trillion operations per second), an installation that will occupy the equivalent of two tennis courts and has cost 600 million dollars in electronics alone. Its power will be equivalent to 10 million times that of a conventional desktop device, which would need 20,000 years to carry out what the new supercomputer does in one day.

The fight to increase the amount of FLOPS on each machine has become a battle between powers. “There is a fight between the US and China to have these better computers,” he explains. “Computers play a determining role in scientific power and grant geostrategic power. That is why there is great interest here, in the EU or Japan to develop them”. And in China. This country “has turned to them. In 2000 they did not have this type. Today, they dominate and have more computers on the list than the US.” Dongarra knows what he is talking about. He was one of the five creators of Top500, the organization that measures the speed of these machines and, above all, he is also the author of the software Linpack back in 1976, the standard program used to measure the power of these super machines.

Jack Dongarra works in the 80s in an image provided by himself.
Jack Dongarra works in the 80s in an image provided by himself.

War is one of the main uses of these devices. “There are a lot of interests in defense and supercomputers in this regard,” he admits. “Like any new technology,” he asserts, “it can be used for both positive and negative ends. The difference between a nuclear reactor and an atomic bomb is very small from a mathematical point of view. Computers can serve to improve our ability to understand things, but in the end, their use is determined by their owners”, he stresses. “Have we developed enough ethics to move things in one direction or another on things like supercomputers or artificial intelligence? Well, I don’t know, that’s a political question.”

Dongarra has developed his entire career in the academic field and in public research centers. “I have acted as a consultant for private companies, but I feel comfortable here. I have never been too interested in working for a company. As a teacher I have the flexibility to work on what I find interesting, and teach it to my students. If I worked in a company I would be obliged to work on their products, I would have someone telling me what is important”.

Do the new generations think like him? Not especially, according to his answer. “I am very concerned about the big companies like Google, Facebook, Microsoft or Amazon in the US. They have an immense amount of resources to develop their products, their financing is practically infinite. So they can take all the talent with astronomical salaries and stock options that I can’t offer, and that makes it really hard to find the right people.”

And after Turing, retirement. Or something similar. “Retirement is a curious word”, he laughs when asked. “I’m stopping teaching, yes”, he explains, “I’m getting rid of the most boring aspects, but I keep my office, I’ll continue researching, I’ll be able to count on students, I’ll apply for scholarships…”. In short: “No, I’m not going to dedicate myself to playing golf.”

You can follow THE COUNTRY TECHNOLOGY in Facebook Y Twitter or sign up here to receive our weekly newsletter.

Exclusive content for subscribers

read without limits



LEAVE A REPLY

Please enter your comment!
Please enter your name here