Reprinted from the "AI Technology Review" public account
Author | Ailleurs
Edit | Chen Caixian
Just now, the highest award in the field of computing in 2021 - the Turing Awards was announced! American computer scientist Jack J. Dongarra was awarded the award for his outstanding achievements in the field of high-performance computing.
According to ACM, Dongarra's algorithms and software have driven the development of high-performance computing, which has had a significant impact on multiple areas of computing science such as artificial intelligence and computer graphics.
He has made pioneering contributions to numerical algorithms and libraries that have enabled high-performance computing software to keep up with more than four decades of exponential hardware updates.
Known as the "Nobel Prize in Computing," the Turing Prize was established in 1966 by the American Computer Society (ACM) in honor of Alan Turing, a pioneer in computer science.M in the world. Turing), one or two scientists who have made significant contributions to the field of computing are selected each year with a reward of $1 million, fully sponsored by Google.
1. Who is Jack Dongarra?
Born on July 18, 1950, Jack J. Dongarra has been a Distinguished Professor in the Department of Electrical Engineering and Computer Science at the University of Tennessee since 1989 and a distinguished researcher in the Computer Science and Mathematics Division of Oak Ridge National Laboratory. Since 2007 he has also been a Turing Fellow at the School of Mathematics at the University of Manchester and an adjunct professor in the Department of Computer Science at Rice University.
His academic experiences are as follows:
He received a bachelor's degree in mathematics from Chicago State University in 1972
He received a master's degree in computer science from the Illinois Institute of Technology in 1973
He received his Ph.D. in Applied Mathematics from the University of New Mexico in 1980 under the tutelage of Cleve Moler, a fellow of the American Academy of Engineering
After graduating with a Ph.D. and joining the University of Tennessee, he worked at Argonne National Laboratory.
Looking back on Jack J. Dongarra's research career, it's been a lot of time: he has won the IEEE Computer Pioneer Award, the SIAM/ACM Computational Science and Engineering Award, and the ACM/IEEE Kennedy Award, and is also an ACM Fellow, IEEE Fellow, SIAM Fellow, AAAS Fellow, ISC Fellow, and IETI Fellow. Fellow Grand Slam.
He is also a Fellow of the National Academy of Engineering and a Foreign Fellow of the Royal Society.
Looking at Google Scholar, his citations exceeded 110,000 and his H-index exceeded 130:
2. His research contributions
According to the ACM website, Dongarra has led the world of high-performance computing through its contributions to efficient numerical algorithms, parallel computing programming mechanisms, and performance evaluation tools for linear algebra operations.
For nearly four decades, Moore's Law has led to exponential growth in hardware performance. At the same time, while most software has failed to keep pace with these hardware advances, high-performance numerical software has done so — thanks in large part to Dongarra's algorithms, optimization techniques, and production-quality software implementations.
These contributions lay the framework for scientists and engineers to make important discoveries and game-changing innovations in areas such as big data analytics, healthcare, renewable energy, weather forecasting, genomics, and economics. Dongarra's work also helps facilitate leapfrog computer architecture and supports the revolution in computer graphics and deep learning.
Dongarra's main contribution was the creation of open source software libraries and standards that use linear algebra as an intermediate language that can be used by a variety of applications. These libraries are 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.
As a leading researcher in high-performance computing, Dongarra leads the field in convincing hardware vendors to optimize these approaches and convincing software developers to target their open source libraries in their work. Ultimately, these efforts resulted in linear algebra-based software libraries enabling near-universal, high-performance scientific and engineering computing on machines ranging from laptops to the world's fastest supercomputers. These libraries are critical to the development of the field – enabling increasingly powerful computers to solve computationally challenging problems.
Gabriele Kotsis, chairman of ACM, said:
"In addition to the interest in breaking new records, high-performance computing has been a major tool for scientific discovery. HPC innovation has also spread to many different computing areas, driving our entire field forward. Jack Dongarra has played a central role in guiding the successful development of this field. His pioneering work dates back to 1979 and he remains one of the most important and active leaders in HPC today. His career undoubtedly reflects the Turing Awards' recognition of "significant contributions of enduring importance.""
Google's Jeff Dean also commented:
"Jack Dongarra's work fundamentally changed and advanced scientific computing. His in-depth and important work at the heart of the world's most frequently used value libraries is the foundation of all areas of scientific computing, helping to advance developments in everything from drug discovery to weather forecasting, aerospace engineering, and dozens of other fields, and his focus on characterizing a wide range of computers has brought significant advances to computer architectures that make them ideal for numerical computing."
For more than four decades, Dongarra has been a principal implementer or principal investigator of multiple libraries such as LINPACK, BLAS, LAPACK, ScaLAPACK, PLASMA, MAGMA, and SLATE. These libraries are written for single processors, parallel computers, multi-core nodes, and multiple GPUs per node. His software libraries are almost universally used to perform high-performance scientific and engineering computing on machines ranging from laptops to the world's fastest supercomputers.
These libraries embody many profound technological innovations, such as:
Auto-tuning: From his project, which won the 2016 Supercomputing Conference Time Testing Award atlas, Dongarra pioneered a way to automatically find algorithmic parameters that produce linear algebraic kernels that are close to optimal efficiency, often superior to vendor-supplied code.
Mixed-precision arithmetic: In his paper "Exploiting the Performance of 32-bit Floating Point Arithmetic in Obtaining 64-bit Accuracy," which he was accepted at the SC Conference in 2006, Dongarra pioneered the use of the multiple precisions of floating-point arithmetic to deliver accurate solutions faster. Recent test demonstrations of the HPL-AI benchmark, which achieve unprecedented levels of performance on the world's top supercomputers, have played an important role in machine learning applications that achieve unprecedented levels of performance on the world's top supercomputers.
Batch Computing: Dongarra pioneered the paradigm of dividing computations for large dense matrices into independent and parallel computations, often used for simulation, modeling, and data analysis. Based on his 2016 paper "Performance, design, and autotuning of batched GEMM for GPUs," Dongarra led the development of the "batch BLAS standard" for such computations and applied to the software libraries MAGMA and SLATE.
Working with many international academics in these efforts, Dongarra has always played a role as an innovation driver by continuously developing new technologies to maximize performance and portability, while using state-of-the-art technology to maintain numerically reliable results.
Other research he leads includes the Messaging Interface (MPI), the de facto standard for portable messaging in parallel computing architectures, and the Performance API (PAPI), which provides an interface that allows performance from components to be collected and synthesized from heterogeneous systems. The standards he helped create, such as MPI, LINPACK benchmarks, and the Top500 supercomputer list, underpin computational tasks from weather forecasting to climate change to analyzing data from large physical experiments.
Reference Links:
https://amturing.acm.org
Due to the trial of the WeChat public account out-of-order push, you may no longer be able to receive pushes from Mozi Salon on time. In order not to be separated from Xiaomo, please set "Mozi Salon" to a star account and often click "Watching" in the lower right corner at the end of the article.
In order to provide better service, "Mozi Salon" has staff to provide special answers on various matters:
Mozi Salon is a large-scale public welfare science popularization forum named after the Chinese sage "Mozi", hosted by the Shanghai Research Institute of the University of Science and Technology of China, and co-organized by the Xinchuang Alumni Foundation of the University of Science and Technology of China, the Education Foundation of the University of Science and Technology of China, the Pudong New Area Science and Technology Association, the China Association for Science and Technology and the Pudong New Area Science and Technology and Economic Committee.
Mozi is a famous thinker and scientist in ancient times on the mainland, and his ideas and achievements are the embodiment of the early scientific buds of the mainland, and the establishment of the "Mozi Salon" aims to inherit and carry forward the scientific tradition, build a social atmosphere advocating science, enhance the scientific literacy of citizens, and advocate and carry forward the spirit of science. The object of popular science is the general public who loves science, has the spirit of exploration and curiosity, and we hope that the public with the equivalent academic strength of middle school and above can understand and appreciate the most cutting-edge scientific progress and scientific ideas in the world.
About "Mozi Salon"