By studying the theoretical limits of how light can be used to perform computation, Cornell researchers have uncovered new insights and strategies for designing energy-efficient optical computing systems.<\/p>\n
The research, published in Nature Communications<\/em>, addresses one of the key challenges to engineering computers that run on light instead of electricity: making those devices small enough to be practical. Just as algorithms on digital computers require time and memory to run, light-based systems also require resources to operate, including sufficient physical space for light waves to propagate, interact and perform analog computation.<\/p>\n Lead authors Francesco Monticone, associate professor of electrical and computer engineering, and Yandong Li, Ph.D. \u201923, postdoctoral researcher, revealed scaling laws for free-space optics and photonic circuits by analyzing how their size must grow as the tasks they perform become more complex.<\/p>\n \u201cOptical computing can be powerful, especially in terms of energy efficiency, but if you need an optical setup that\u2019s as large as an entire room to accomplish a meaningful AI inference task, such as image classification, then your optical computer isn\u2019t very practical. Keep in mind that photons are much harder to confine in small spaces than electrons,\u201d Monticone said. \u201cWe wanted to understand the fundamental tradeoffs between task complexity, performance, and minimum physical size, and then find ways to design optical inference systems that use the available space optimally.\u201d<\/p>\n To address this, the researchers found inspiration in a deep-learning technique called \u201cneural pruning,\u201d which safely removes, or prunes, redundant parameters with little impact on performance.<\/p>\n \u201cWe specifically analyzed the connectivity pattern of these optical devices \u2013 how light waves overlap and interact in the entire device,\u201d Li said. \u201cThen we developed optics-specific pruning methods grounded in wave physics to penalize the overlapping of light waves. This allowed us to simplify the network considerably with minimal loss in accuracy.\u201d<\/p>\n Using this technique, the researchers found that an optical computing system performing the same task could be 1% to 10% the size of its conventional counterpart. To put their findings in perspective, the researchers estimated how large an optical computer would need to be to perform the linear operations in large language models such as ChatGPT \u2013 more specifically, at a scale of 100 billion to 2 trillion parameters. They found that a free-space optical setup could, in principle, perform computations at this scale in a device roughly 1 centimeter thick. To approach this theoretical limit, emerging optical devices such as ultra-thin metasurfaces and spaceplates could be promising candidates, as the researchers noted in an earlier paper.<\/p>\n The study also revealed a trend of diminishing returns in inference accuracy as the optical device becomes larger, meaning that for some applications it is better to strike a balance between device size and task performance, according to the researchers.<\/p>\n While fully optical computers remain a challenging long-term goal, Li and Monticone see more immediate applications in hybrid systems, where light handles fast, energy-intensive linear operations, and electronics provides nonlinear functions, branching logic, decision-making and general-purpose programmability.<\/p>\n \u201cThere are limitations other than size that make me personally skeptical about whether optical computers will really replace, or drastically accelerate, things like GPUs,\u201d Monticone said. \u201cBut for many applications such as imaging and computing in resource-limited edge scenarios, optics could work extremely well, and we show that space is not necessarily the bottleneck some people feared.\u201d<\/p>\n <\/p>\n For more information:<\/p>\n