Peking University Unveils Groundbreaking Analog Computing Chip

Researchers at Peking University have developed a high-precision, scalable analog matrix computing chip that significantly outperforms current top digital processors, such as Nvidia's GPUs, in both computational speed and energy efficiency. This breakthrough, detailed in the October 13, 2025, issue of Nature Electronics, leverages resistive random-access memory (RRAM) to address complex scientific computations with unprecedented precision and scalability.

Analog computing processes information through continuous electrical signals, offering potential advantages in speed and energy efficiency over digital computing, which relies on discrete binary code. However, analog systems have historically faced challenges in precision and scalability, limiting their practical applications. Resistive random-access memory (RRAM) is a type of non-volatile memory that stores data by changing the resistance across a dielectric solid-state material. RRAM's ability to perform in-memory computing makes it a promising candidate for analog computing applications.

The research team, led by Associate Researcher Sun Zhong from the Institute for Artificial Intelligence at Peking University, collaborated with the School of Integrated Circuits to develop this innovative chip. Their approach combined novel devices, original circuits, and classical algorithms to build a high-precision, scalable analog matrix solver using resistive memory arrays. For the first time, they achieved 24-bit fixed-point accuracy in analog computing, improving the precision of traditional analog computing by five orders of magnitude.

In performance tests, the chip demonstrated remarkable capabilities. When computing the inverse of a 32×32 matrix, the chip's computational prowess outstripped the single-core performance of high-end GPUs. As the problem size escalated to a 128×128 matrix, the chip's computational throughput exceeded that of top-tier digital processors by over 1,000 times. Additionally, at an equivalent level of precision, the chip's energy efficiency was over 100 times greater than that of traditional digital processors.

The researchers emphasized the significance of their achievement. "Precision has long been the central bottleneck of analogue computing," the authors stated in their paper. Study author Sun Zhong noted, "How to achieve both high precision and scalability in analogue computing, thereby leveraging its inherent advantages for modern computing tasks, has been a 'century-old problem' plaguing the global scientific community."

This development has significant implications across various fields. The chip's high throughput and energy efficiency make it well-suited for energy-intensive and high-throughput applications such as artificial intelligence (AI) and future 6G communication systems. Its ability to compute directly within memory hardware bypasses the traditional energy drain of data transfer between CPU and memory, making it ideal for complex scientific computations.

The chip was built using commercial manufacturing processes, indicating the potential for mass production. Researchers plan to develop larger, more integrated versions for tackling even more complex computations.

Peking University's development of a high-precision, scalable analog matrix computing chip represents a significant breakthrough in computing technology. By overcoming longstanding challenges in analog computing, this innovation paves the way for more efficient and powerful computing solutions across various scientific and technological domains.