TechNews, April 12th – According to reports from the media, Tsinghua University has pioneered a distributed wide-light computing architecture and developed the large-scale interference-diffraction heterogeneous integrated chip “Taichi”, achieving a general intelligent computing capability of 160 TOPS/W.

This research achievement discards the traditional paradigm of electronic deep computing and instead, follows the path of photons, exploring new inspirations, architectures, and pathways for high-performance computing.

With the flourishing development of artificial intelligence, intelligent optical computing, as an emerging computing mode, has demonstrated performance and potential far beyond silicon-based electronic computing in the post-Moore era.

However, the biggest challenge lies in the fact that the advantages of optical computing are trapped in inappropriate electronic architectures, limiting the scale of computation and unable to support the urgent need for complex large-scale intelligent computing with high computing power and efficiency.

The computing efficiency of the Taichi optical chip directly surpasses existing intelligent chips by 2 to 3 orders of magnitude, providing computational support for ultra-fast intelligent analysis of billion-pixel large scenes, training inference of billion-parameter large models, and milliwatt-level low-power autonomous intelligent unmanned systems.