The American Computer Association (ACM) awarded JackJ the Turing Award of 202 1. Dongarra, a 7 1 year-old professor in the Department of Electrical Engineering and Computer Science, University of Tennessee, USA, commended his pioneering contributions in scale optimization algorithm and special tool library, which enabled high-performance computing mobile phone software to keep up with the exponential hardware configuration upgrade in the past 40 years.
According to the detailed introduction of ACM, Dongarra's computing method and mobile phone software have promoted the development trend of high-performance computing, and caused important harm to many computing science industries from artificial intelligence technology to computer image processing.
Turing Award is an honorary award awarded by the American Computer Association in 1966. It is dedicated to rewarding people who have made great contributions to computer work, and is called "Nobel Prize in computer field". The Turing Award is $654.38+$00,000, funded by Google. This award is named after Aram. British mathematician Turing, who showed the limits of basic mathematics and calculation.
Jack. Tangala was born in 1950, 18 in July, and obtained a bachelor's degree certificate in mathematics from Chicago State University in 1972 in June. Master of Computer Science, Illinois Institute of Technology1973; 1980 obtained a doctorate in applied mathematics from the University of New Mexico, and studied under CleveMoler, the founder of Matlab and an academician of the National Academy of Engineering.
Since 1989, Dongarra has been a permanent professor in the Department of Electrical Engineering and Computer Science of the University of Tennessee, and an outstanding researcher in the Department of Computer Science and Mathematics of Oak Ridge National Laboratory. Since 2007, he has been a Turing researcher at the University of Manchester, England, and a part-time expert professor at the Department of Computer Science at Rice University.
Dongarra has won many awards in his scientific research career, including IEEE Computer Pioneer Award, SIAM/ACM Computing Science and Engineering Project Award and ACM/IEEEKennedy Award. He is a member of ACM, IEEE, SIAM, AAAS, ISC and IETI. In addition, he is an academician of the National Academy of Engineering and a distinguished foreign member of the Royal Society.
Dongarra promotes global high-performance computing based on efficient scalar optimization algorithm, parallel processing computing programming system and performance evaluation equipment. In recent forty years, Moore's Law has made the performance of hardware configuration increase exponentially. At the same stage, although most mobile phone software can't keep up with the development of this hardware configuration, high-performance standard mobile phone software is ready, which is largely due to Dongarra's optimization algorithm, development technology and software realization of manufacturing quality.
This dedication has established a framework for scientists and technical engineers to make major discoveries and make independent innovations in the fields of data analysis, medical care, renewable resources, weather trends, molecular biology and socio-economics. Donggara's work is also conducive to promoting the leap-forward development of computer system architecture, which is suitable for the revolution of computer image processing and deep learning.
Dongarra's key contributions also include the establishment of open source project libraries and specifications, which can be used by various applications by using line generation as a front and intermediate language expression. This library is written for single CPU, parallel processing computer, multi-core connection point and several GPUs per node. The library also introduces many key independent innovations, including automatic adjustment, mixed precision arithmetic and batch command calculation.
As a pioneer of high performance computing, Dongarra led the hardware configuration dealers in this industry to upgrade this method, and persuaded the mobile phone software developers to take his open source system library as the overall goal in their work. Finally, from notebooks to fast computers all over the world, Dongarra's diligence has made the software library based on line generation widely used in high-performance science and engineering project calculation. This kind of library is particularly important for the future development of this industry, which enables computers with more and more powerful functions to deal with challenging computing problems.
GabrieleKotsis, the current chairman of ACM, said: "Nowadays, faster extraordinary computers have become headlines in the news media, and people are interested in the amazing pioneering work of performing trillions of calculations in one second. But apart from getting rid of the interest in recording, high performance computing (HPC) has always been the first special tool for scientific discovery. The independent innovation of HPC has also flocked to many different computing industries, which has promoted the development trend of all industries. JackDongarra has played a major role in guiding the successful development trend in this field. His pioneering work can be traced back to 1979, and he is still one of the most important and active managers in HPC community. His career development undoubtedly shows the Turing Award's recognition of' outstanding contribution with long-term necessity'. "
JeffDean, a senior researcher at Google and senior vice president of Google's scientific research and Google's physical and mental health, said: "JackDongarra's work has changed and promoted the development trend of scientific computing from the source. The key deep-seated work he has done in the most frequently used standard library in the world today is the scientific calculation of all walks of life, which has helped promote the development trends of dozens of industries from drug discovery to weather forecast and aerospace engineering, and has helped promote the development trends of dozens of industries from drug discovery to weather forecast. He is committed to showing that general-purpose computers have made important developments for computer system architecture, especially suitable for scalar computing. " Donggara will attend the ACM Annual Awards Dinner in San Francisco on June 1 1 2022, and be awarded the ACM Turing Award at the dinner.
For more than forty years, Dongarra has been a major participant or president researcher in several libraries, such as LINPACK, BLAS, LAPACK, ScaLAPACK, PLASMA, MAGMA and SLATE. This library is written for single CPU, parallel processing computer, multi-core connection point and several GPUs per node. From notebooks to fast computers all over the world, his software library is widely used in this kind of equipment to realize high-performance scientific and engineering project calculation.
This library shows many in-depth technological innovations, such as: automatic tuning: according to his new ATLAS project, this project won the time detection award at the 20 16 World Extraordinary Computing Exchange Conference. Dongarra opened up a fully automatic method to find the main parameters of the optimization algorithm, which led to the core of line generation close to the best working efficiency, usually better than the coding brought by dealers.
Mixed precision arithmetic: in the paper of the 2006 World Supercomputing Conference, Dongarra used the various precisions of floating-point arithmetic for the first time to give an accurate solution quickly. As recently demonstrated in the HPL-AI benchmark, this work has stimulated the key functions in the application of artificial neural network, and achieved unprecedented performance level in the world's top extraordinary computers.
Large-scale calculation: Dongarra pioneered the case of dividing large and medium-sized aggregation and drainage matrix calculation (usually used for simulation, model and data statistical analysis) into several task calculations. This daily task block can be calculated separately and has high concurrency. According to his graduation thesis "Performance, Design and Automatic Tuning of Batchedgemmforgpus" published on 20 16, leading cadres of Dongarra developed batch command BLAS specifications for this kind of calculation, which also occurred in software libraries MAGMA and SLATE.
Dongarra has conducted international exchanges with many people on the above work. He plays the role of driving force of independent innovation from beginning to end, maximizing performance and portability according to the continuous development of technology, while applying the most modern technology to maintain reliable data results.
In addition, he also led cadres to develop Message Passing Socket (MPI), which is the de facto standard for life cycle message passing on parallel processing computing architecture. And its performance API(PAPI), which ensures a socket that allows performance to be collected and generated from components of heterogeneous operating systems. The specifications he helped establish, such as MPI, LINPACK standard and Top500 computer catalog, support daily tasks ranging from weather trends to climate problems to analyzing data information of large-scale physical experiments.