Ryan Hamerly

Liane Bernstein, Alexander Sludds, Christopher Panuski et al.

As deep neural networks (DNNs) grow to solve increasingly complex problems, they are becoming limited by the latency and power consumption of existing digital processors. For improved speed and energy efficiency, specialized analog optical and electronic hardware has been proposed, however, with limited scalability (input vector length $K$ of hundreds of elements). Here, we present a scalable, single-shot-per-layer analog optical processor that uses free-space optics to reconfigurably distribute an input vector and integrated optoelectronics for static, updatable weighting and the nonlinearity -- with $K \approx 1,000$ and beyond. We experimentally test classification accuracy of the MNIST handwritten digit dataset, achieving 94.7% (ground truth 96.3%) without data preprocessing or retraining on the hardware. We also determine the fundamental upper bound on throughput ($\sim$0.9 exaMAC/s), set by the maximum optical bandwidth before significant increase in error. Our combination of wide spectral and spatial bandwidths in a CMOS-compatible system enables highly efficient computing for next-generation DNNs.

Ronald Davis, Zaijun Chen, Ryan Hamerly et al.

Edholm's Law predicts exponential growth in data rate and spectrum bandwidth for communications and is forecasted to remain true for the upcoming deployment of 6G. Compounding this issue is the exponentially increasing demand for deep neural network (DNN) compute, including DNNs for signal processing. However, the slowing of Moore's Law due to the limitations of transistor-based electronics means that completely new paradigms for computing will be required to meet these increasing demands for advanced communications. Optical neural networks (ONNs) are promising DNN accelerators with ultra-low latency and energy consumption. Yet state-of-the-art ONNs struggle with scalability and implementing linear with in-line nonlinear operations. Here we introduce our multiplicative analog frequency transform ONN (MAFT-ONN) that encodes the data in the frequency domain, achieves matrix-vector products in a single shot using photoelectric multiplication, and uses a single electro-optic modulator for the nonlinear activation of all neurons in each layer. We experimentally demonstrate the first hardware accelerator that computes fully-analog deep learning on raw RF signals, performing single-shot modulation classification with 85% accuracy, where a 'majority vote' multi-measurement scheme can boost the accuracy to 95% within 5 consecutive measurements. In addition, we demonstrate frequency-domain finite impulse response (FIR) linear-time-invariant (LTI) operations, enabling a powerful combination of traditional and AI signal processing. We also demonstrate the scalability of our architecture by computing nearly 4 million fully-analog multiplies-and-accumulates for MNIST digit classification. Our latency estimation model shows that due to the Shannon capacity-limited analog data movement, MAFT-ONN is hundreds of times faster than traditional RF receivers operating at their theoretical peak performance.

Kfir Sulimany, Sri Krishna Vadlamani, Ryan Hamerly et al.

Secure multiparty computation enables the joint evaluation of multivariate functions across distributed users while ensuring the privacy of their local inputs. This field has become increasingly urgent due to the exploding demand for computationally intensive deep learning inference. These computations are typically offloaded to cloud computing servers, leading to vulnerabilities that can compromise the security of the clients' data. To solve this problem, we introduce a linear algebra engine that leverages the quantum nature of light for information-theoretically secure multiparty computation using only conventional telecommunication components. We apply this linear algebra engine to deep learning and derive rigorous upper bounds on the information leakage of both the deep neural network weights and the client's data via the Holevo and the Cramér-Rao bounds, respectively. Applied to the MNIST classification task, we obtain test accuracies exceeding $96\%$ while leaking less than $0.1$ bits per weight symbol and $0.01$ bits per data symbol. This weight leakage is an order of magnitude below the minimum bit precision required for accurate deep learning using state-of-the-art quantization techniques. Our work lays the foundation for practical quantum-secure computation and unlocks secure cloud deep learning as a field.

Lian Zhou, Kaiwen Xue, Yun-Jhu Lee et al.

High-performance computing underpins modern artificial intelligence (AI), enabling foundation models, real-time inference and perception in autonomous systems, and data-intensive scientific simulations. Recent advances in quantization techniques utilizing low-precision computation without degrading model accuracy, create new opportunities for analog photonic computing characterized by ultra-high clock rates and low energy consumption. Here we propose and demonstrate a coherent homodyne integrated circuit capable of general matrix multiplication (GEMM) with aggregate throughput that exceeds 1,000 TOPS (tera-operations per second), enabled by massive on-chip optical fanout and parallelism. By leveraging time multiplexing, the required modulator count is reduced from O($N^2$) to O(N), allowing dense integration of record-scale 256 $\times$ 256 homodyne units (each <0.0064 $mm^2$) within a single reticle. We employ wafer-scale fabricated 64 thin-film lithium niobate (TFLN) transmitters (each over 40-GHz bandwidth with propagation loss of 0.2 dB/cm) to encode data and chip-to-chip coupled to Si/SiN computing circuits (64 channels). Our system achieves up to 7-bit computational accuracy across 8 $\times$ 8 parallel channels at record computing clockrate 120 Gbaud/s, and 6-bit statistical accuracy across 256 $\times$ 100 channels at 20-128 Gbaud/s, representing a total throughput of 1,000-6,000 TOPS. Massive parallelism amortizes the optoelectronic (OE) conversion to allow 330-TOPS/W efficiency using foundry-available packaging technology. The system throughput is benchmarked with Qwen2.5-0.5 billion parameter models that generate accurate tokens. High throughput and energy efficiency establish a near-term pathway toward light-based accelerators for large-scale training and low-latency inference from datacenters to edges, accelerating new models toward artificial general intelligence.