China develops world's first memristor neuromorphic dynamics chip based on controllable in-memory computing
Original title:我国研制全球首款神经动力学芯片,首次将单步运算时延压缩至 2.12 毫秒
AI summary
The School of Integrated Circuits at Peking University, together with the Shanghai Institute of Microsystem and Information Technology of the Chinese Academy of Sciences, released the world's first memristor neuromorphic dynamics chip based on controllable in-memory computing, compressing single-step computation latency to 2.12 milliseconds for the first time. The chip uses a 40-nanometer process, with an in-memory computing array and peripheral circuits occupying a total area of 0.28 square millimeters, operating at 50 MHz and requiring only 9 pipeline stages for single-step integration. In tasks such as cortical reconstruction, it is 50 to 478 times faster than current GPUs, breaking through the real-time computing bottleneck of neuromorphic dynamics. The results were published in Science on July 3.
Key Takeaways
- The School of Integrated Circuits at Peking University,together with the Shanghai Institute of…
- 12 milliseconds for the first time.
- The chip uses a 40-nanometer process,with an in-memory computing array and peripheral circuits…
- 28 square millimeters,operating at 50 MHz and requiring only 9 pipeline stages for single-step…
- In tasks such as cortical reconstruction,it is 50 to 478 times faster than current GPUs,breaking…
Quick Facts
- Published
- Updated
- Source
- IT之家·人工智能
- Category
- Papers & Research
- AI score
- 74
全球首款神经动力学芯片将单步时延压至2.12毫秒,首次实现实时计算,比GPU快50倍以上,这是神经拟态芯片落地的关键一步,做实时仿真和边缘AI的可以留意。
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: China develops world's first memristor neuromorphic dynamics chip based on controllable in-memory computing. July 4, 2026. https://new.nat6.net/en/news/cmr5wizx2067eslc7xynzcf9y/