Virtualizing a Post-Moore’s Law Analog Mesh Processor: The Case of a Photonic PDE Accelerator

Abstract

Innovative processor architectures aim to play a critical role in future sustainment of performance improvements under severe limitations imposed by the end of Moore’s Law. The Reconfigurable Optical Computer (ROC) is one such innovative, Post-Moore’s Law processor. ROC is designed to solve partial differential equations in one shot as opposed to existing solutions, which are based on costly iterative computations. This is achieved by leveraging physical properties of a mesh of optical components that behave analogously to lumped electrical components. However, virtualization is required to combat shortfalls of the accelerator hardware. Namely, 1) the infeasibility of building large photonic arrays to accommodate arbitrarily large problems, and 2) underutilization brought about by mismatches in problem and accelerator mesh sizes due to future advances in manufacturing technology. In this work, we introduce an architecture and methodology for light-weight virtualization of ROC which exploits advantages borne from optical computing technology. Specifically, we apply temporal and spatial virtualization to ROC and then extend the accelerator scheduling tradespace with the introduction of spectral virtualization. Additionally, we investigate multiple resource scheduling strategies for a system-on-chip (SoC)-based PDE acceleration architecture and show that virtual configuration management offers a speedup of approximately 2 ×. Finally, we show that overhead from virtualization is minimal, and our experimental results show two orders of magnitude increased speed as compared to microprocessor execution while keeping errors due to virtualization under 10%.

Demultiplexing OAM beams via Fourier optical convolutional neural network

“…[108][109][110] Furthermore, these topologically safeguarded modes can be utilized in long-distance free-space communication links, capitalizing on the non-diffracting and self-healing properties of OAM. [111][112][113][114][115][116] In this previous work, we proposed a swift multiplexing and demultiplexing scheme, supplemented by an 'intelligent' active demultiplexing mechanism that facilitates adaptivity to the transmission medium. The overarching goal is to achieve GHz-speed optical communication suitable for underwater deployment.…”

Exciton strain mapping in 2D TMDCs for sensing and photodetection

“…Here we performed an extensive study on nonintegrated WS 2 photodetectors and integrated WS 2 photodetectors. [35][36][37][38][39][40][41][42][43][44][45] Also, it has been demonstrated that the exciton bandgap in TMDCs can be shifted and tuned as an effect of strain. 37,[46][47][48][49] It has been experimentally demonstrated that strain is induced in TMDCs integration on photonic circuits due to the structural profile owing to bandgap tuning.…”

Power conservation version of application-specific photonic integrated circuit

“…[54][55][56][57][58][59][60][61][62][63][64][65][66][67][68][69][70][71][72][73] Solving Partial Differential Equations (PDEs) is still one of the essential tasks in everyday life in many fields, from engineering to financial services. [74][75][76][77][78][79][80][81] There are some technologies for lightweight virtualization of analog computers that exploit advantages borne from optical computing technology, [82][83][84][85][86] which apply temporal and spatial virtualization to the analog solver and then extend the accelerator scheduling Trade-space with the introduction of spectral virtualization. Digital computing had a significant impact on the processing of this task, allowing to solve PDE models in a short time.…”