The State Key LAB of Processors Won the Most Influential Paper Award at ASPLOS 2024

Date: May 15, 2024
On April 30, 2024, ASPLOS 2024 conference, an A-class international top-tier conference, announced that the paper "DianNao: a small-footprint high-throughput accelerator for ubiquitous machine-learning", has been awarded the Influential Paper Award. The DianNao paper was published ten years ago at ASPLOS 2014 by State Key Laboratory of Processors (SKLP), Institute of Computing Technology (ICT), Chinese Academy of Sciences (CAS).

The award recognizes historical ASPLOS papers that have had major influence on the field. Each year, the ASPLOS program committee nominates award candidates from papers published ten or more conferences ago, and the final selection of the winner(s) is made by the ASPLOS steering committee. The award is formally announced during the ASPLOS'24 banquet on Tuesday, 30 Apr 2024, in San Diego, USA. This is the first time that a Chinese institute won this award, and it is also the first time a Chinese institute won such type of award in the field of computer architecture.


The award-winning DianNao paper was jointly completed by the team of professor Yunji Chen from ICT, and professor Olivier Temam from France Inria. Authors also include Tianshi Chen (currently CEO of Cambricon company), Zidong Du (currently professor at ICT), Ninghui Sun (currently professor at ICT), Jia Wang and Chengyong Wu.

The DianNao paper introduced the world's first deep learning processor architecture, achieving a small size of footprint but with a high throughput deep learning chip with only 3.02mm2 area cost under a 65nm process. This significantly enhances the performance and energy efficiency of deep learning processing by orders of magnitude. The DianNao paper was also awarded the Best Paper Award at ASPLOS 2014.

The DianNao paper pioneered the field of deep learning processors, propelling it to become a hot topic in international academic research and industry. Currently, the DianNao paper has been cited nearly two thousand times (Google Scholar) by researchers across five continents, over thirty countries, and hundreds of institutions. In recent years, the four major architecture conferences (ISCA, MICRO, ASPLOS, HPCA) keep publishing dozens of deep learning processor architecture related papers that cited DianNao each year. Also, after the DianNao paper published, industrial companies, both domestic and international, have devoted to develop deep learning chips, including Google's TPU, NVIDIA's Tensor Core, and Huawei's series of smart chips.

Paper Link: