Search papers, labs, and topics across Lattice.
This research presents a time-encoded analog photonic interposer that facilitates the efficient long-distance transmission of analog signals between chiplets, surpassing the limitations of traditional silicon-photonic links that only handle digital data. By utilizing an analog-to-time converter (ATC) to encode amplitudes as timing intervals, the system achieves significant energy efficiency, demonstrating a 2.04x improvement in energy-delay product (EDP) compared to an 8-bit digital baseline in a fully analog vision pipeline. The interposer maintains high accuracy in vision tasks, achieving 89.87% and 86.15% accuracy on ResNet18 and MobileNetV2, respectively, while generalizing well across different datasets.
Achieving over 2x energy efficiency in analog vision processing could redefine the performance benchmarks for future AI hardware.
This work introduces a time-encoded analog photonic interposer that enables long-distance, high-fidelity transport of analog signals between spatially separated chiplets. Unlike prior silicon-photonic links limited to digital data, the interposer preserves analog information by converting amplitudes into timing intervals using an analog-to-time converter (ATC), transmitting them over a wavelength-division-multiplexed photonic link, and reconstructing values at the receiver without an explicit high-precision ADC/DAC data-converter pipeline. The link instead embeds an implicit 6-bit time-domain quantization and uses a single wavelength per processing element independent of bit precision. Evaluated in a fully analog vision pipeline with an in-pixel computing (IPC) sensor, it achieves a 2.04x energy--delay product (EDP) improvement over an 8-bit digital electrical baseline on the 560x560 Visual Wake Words dataset, with the advantage widening with link length even against a precision-matched 6-bit baseline. The pipeline holds 89.87% and 86.15% accuracy on ResNet18 and MobileNetV2 for VWW and generalizes across CIFAR-10 and ModelNet40 within 2% of the digital baseline.
