Reporting by XinZhiyuan
Edited by David Good Sleepy Layan
【New Zhiyuan Introduction】The 2021 Turing Awards are announced! Jack Dongarra, a pioneer in high-performance computing and one of the founders of the Supercomputing Top500, won the award and received an exclusive $1 million prize.
In the late 1970s, a young researcher at the Aragonese National Laboratory wrote computer code called "Linpack," which also allowed systems later known as supercomputers to run complex mathematical calculations.
In the early 1990s, the researcher and his colleagues used "Linpack" to create a new test to measure the power of supercomputers, that is, to test how many times a supercomputer can perform calculations per second.
This person is Jack Dongarra, winner of the 2021 ACM Turing Award.
Change the way the world calculates
The American Computer Society (ACM) named University of Tennessee Professor Jack J. Dongarra as the 2021 Turing Award winner for his pioneering contributions to numerical algorithms and libraries, and his work has enabled high-performance computing software to keep up with exponential improvements in hardware over the past 40 years.
In addition to a beautiful silver bowl, the award includes a $1 million prize funded entirely by Google, which goes directly to Dongarra.
Dongarra's algorithms and software have driven the development of high-performance computing and have had a significant impact on many areas of computational science, from artificial intelligence to computer graphics.
Known as the "Nobel Prize of Computing", the Turing Award is currently the highest award in the field of computer science in the world, with a prize of $1 million, supported by Google. The prize is named after the British mathematician Alan Turing, the father of modern computers.
Dongarra has led the world of high-performance computing by contributing to efficient numerical algorithms for linear algebra operations, parallel computing programming mechanisms, and performance evaluation tools.
For nearly four decades, Moore's Law has led to exponential growth in hardware performance. During the same period, most software performance failed to keep up with these hardware advances, but high-performance numerical software did. This is largely due to Dongarra's algorithms, optimization techniques and production quality software implementations.
ACM President Gabriele Kotsis said:
"High-performance computing has always been a major tool for scientific discovery. Innovations in high-performance computing have already impacted many different computing areas, driving our entire field forward. Dongarra has played a central role in guiding the success trajectory in this field."
Dongarra's main contribution was the creation of open source software libraries and standards that use linear algebra as an intermediate language for a variety of applications.
These libraries have been written for single processors, parallel computers, multi-core nodes, and multiple GPUs per node. Dongarra's library also introduces a number of important innovations, including auto-tuning, mixed-precision arithmetic, and batch computation.
Dongarra's pioneering work dates back to 1979 and he remains one of the most important and actively engaged leaders in the HPC community to this day. There is no doubt that his career deserves the Turing Award's recognition of "a significant contribution of enduring importance".
Jack Dongarra and his algorithm
Jack Dongarra has been a University Distinguished Professor and Distinguished Investigator at Oak Ridge National Laboratory at the University of Tennessee since 1989. Since 2007, he has also been a Turing Fellow at the University of Manchester in the UNITED Kingdom.
He received his B.S. in Mathematics from Chicago State University, his M.S. in Computer Science from Illinois Institute of Technology, and his Ph.D. in Applied Mathematics from the University of New Mexico.
Dongarra has previously received honors including the IEEE Computer Pioneer Award, the SIAM/ACM Computational Science and Engineering Award, and the ACM/IEEE Kennedy Award.
He is a fellow at ACM, the Institute of Electrical and Electronics Engineers (IEEE), the Industrial and Applied Mathematics Society (SIAM), the American Association for the Advancement of Science (AAAS), the International Conference on Supercomputing (ISC), and the International Institute of Engineering and Technology (IETI), a Fellow of the National Academy of Engineering and a foreign member of the Royal Society.
Now, Dongarra's paper citations have exceeded 110,000.
For the past 50 years, Dongarra has been helping advance numerical algorithms and software, parallel computing programming, and performance benchmarking that are necessary to create hyperscale supercomputers.
Today, from the laptops of ordinary families to the fastest supercomputers in the world, there are shadows of Dongarra's outstanding contributions.
The open source software library created by Dongarra contains many deep technological innovations, such as auto-tuning, mixed-precision computing, and batch computation.
Autotuning:
Dongarra, in the ATLAS project at the 2016 Global Supercomputing Congress, investigated a new method for automatically finding algorithm parameters that generate a linear algebraic kernel that is nearly optimal in efficiency.
This algorithm parameter is generally better than the code performance provided by the manufacturer.
Mixed Precision Arithmetic:
In a paper submitted by Dongarra to the Global Supercomputing Congress in 2006, he proposed to "use 32-bit floating-point algorithms to obtain 64-bit precision performance." He pioneered a way to use multiple times the precision of floating-point calculations to output precise solutions faster.
This research is slowly becoming more fundamental in machine learning applications, most recently the HPL-AL Benchmark, which achieves unprecedented performance on the world's most advanced supercomputing.
Batch Computations:
Dongarra pioneered a paradigm for the operation of splitting large density matrices, which is widely used in simulation, modeling, data analysis and other fields. This paradigm divides the operations of large density matrices into more calculations containing smaller task quantities, which can be operated simultaneously independently of each other.
He has been involved in creating standards, including MPI, LINPACK Benchmark, etc., that lay the foundation for a variety of different computational tasks, such as weather forecasting, climate change, to analyzing data from large-scale physical experiments.
At present, the most authoritative list of supercomputer performance is Top 500, and the evaluation software algorithm behind it is developed by Dongarra.
Even if you don't count the Turing Award you just won, Dongarra is already a success. And the Turing Award is equivalent to adding the brightest stroke to his extraordinary resume.
What do domestic experts think?
After the news of Dongarra's award was announced, many big names in the domestic computer field congratulated and put forward their own views.
Bao Yungang, a researcher at the Institute of Computing Technology of the Chinese Academy of Sciences, said: "The first feeling is a surprise."
Professor Bao is also a big guy in the industry. In addition to his status as a researcher, he is also a professor at the University of Chinese Academy of Sciences and the director of the Center for Advanced Computer Systems Research of the Chinese Academy of Sciences. His main research direction is the cutting-edge research of computer architecture and open source chips.
There are two reasons why Professor Bao is so excited. One is that the field of high-performance computing, which is so important, has finally turned the Turing Award. Dongarra's award this time can be said to be a late Turing Award in the field of high-performance computing.
Why is it "late"?
In the last century, there was also a great scientist who made outstanding contributions in the field of high-performance computing, Seymour Cray.
In 1958, Cray designed and built the world's first transistor-based supercomputer, which became an important milestone in the history of computer development. At the same time, it has also made a significant contribution to the production of RISC high-end microprocessors.
Unfortunately, on September 22, 1996, Cray died of serious injuries in an accidental car accident and died on October 5 of the same year at the age of 71. If Professor Cray were still alive, he would probably have won the Turing Award for his achievements.
Another reason is that Jack Dongarra is actually an old friend of China's high-performance computing community. Before the epidemic, he had to come to China many times a year, shuttling through various supercomputing centers in China. Many domestic industry leaders have a deep friendship with Dongarra.
Dongarra was interviewed by a small Chinese reporter
In addition to the contribution of Dongarra mentioned above, Professor Bao also added another point, that is, the birth of Matlab.
Professor Bao said, "What is the origin of Matlab? Professor Moore of the University of New Mexico developed several pieces of software in the 1970s, one of which was Linpack. Later, in order to facilitate teaching, he wrote a gadget called Matlab to encapsulate them."
Dongarra's supervisor is Professor Moore, who did the same work as developing LINPACK during his Ph.D. So in a sense, the emergence of Matlab is inseparable from Dongarra.
Professor Bao concluded: "If you do something difficult, you will gain something."
Congratulations again to Professor Dongarra, turing award in the field of high-performance computing, although late but late!
Resources:
https://amturing.acm.org/
https://www.nytimes.com/2022/03/30/technology/turing-award-jack-dongarra.html
https://www.zhihu.com/question/525065780/answer/2415726088
You can invite the group by following the "New Zhiyuan" WeChat public account, adding a small assistant (aiera2015_2), and the community has daily information sharing, resource exchanges, etc.
0.0MB