Photonic Memory Interconnect Modules for Scalable Bandwidth
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional computing systems face limitations in scaling memory capacity and bandwidth, particularly for data-intensive applications like deep learning, due to parasitic impedance in silicon interposers and the need for memory chips to be positioned close to processors, which restricts memory bandwidth and capacity.
Innovation Solution
The development of photonic communication platforms using photonic modules with optical distribution networks and waveguides, which are immune to parasitic impedance, allowing for scalable memory capacity and bandwidth by enabling optical communication between memory and processor modules without the need for close proximity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If memory chips are positioned close to processors using silicon interposers, then electrical communication is enabled, but parasitic impedance increases and memory bandwidth is restricted
Solution Approach 1:
The patent replaces electrical communication (mechanical/electrical system) with optical communication using photonic modules, waveguides, and optical interconnects. This substitution eliminates parasitic impedance inherent in electrical systems while maintaining reliable communication between processors and memory chips, even at greater distances.
Solution Approach 2:
The patent introduces photonic modules as intermediary components between processors and memory chips. These modules contain optical-to-electrical and electrical-to-optical converters, waveguides, and optical switches that mediate communication, allowing signals to travel optically through the interposer without suffering from electrical parasitic impedance.
2Reliability
If memory chips are positioned close to processors, then electrical communication is achieved, but memory capacity and bandwidth scaling is limited
Solution Approach 1:
The patent implements reconfigurable optical interconnects using optical switches and programmable photonic modules. This dynamic capability allows the system to adaptively reconfigure communication paths and allocate bandwidth flexibly, enabling memory capacity and bandwidth to scale according to application requirements rather than being fixed by physical proximity constraints.
Solution Approach 2:
The patent transitions from two-dimensional electrical trace routing on silicon interposers to three-dimensional optical waveguide routing. Optical waveguides can be routed through multiple layers and dimensions within the photonic module, enabling more flexible and scalable communication paths that are not constrained by the planar limitations of electrical interconnects.
3Manufacturing precision
If different photomask sets are used for photonic module fabrication, then manufacturing precision is improved, but manufacturing cost increases
Solution Approach 1:
The patent designs photonic modules with standardized, universal photomask sets that can fabricate multiple types of photonic modules (e.g., optical-to-electrical converters, electrical-to-optical converters, optical switches, waveguides) using the same masking process. This universality maintains manufacturing precision across different module types while significantly reducing fabrication costs by eliminating the need for separate photomask sets for each module type.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This solution enables significant scaling of memory capacity and bandwidth beyond conventional limits, supporting data-intensive applications by allowing flexible computer architectures and reducing manufacturing costs through the use of common photomask sets for photonic module fabrication.
Implementation Method 1
a first optical waveguide optically coupling the optical distribution network to a first neighboring photonic module
Implementation Method 2
the optical distribution network is configured to selectively place the first neighboring photonic module in optical communication with the second neighboring photonic module
Data Source
AI summary
Described herein are photonic communication platforms that can overcome the memory bottleneck problem, thereby enabling scaling of memory capacity and bandwidth well beyond what is possible with conventional computing systems. Some embodiments provide photonic communication platforms that involve use of photonic modules. Each photonic module includes programmable photonic circuits for placing the module in optical communication with other modules based on the needs of a particular application. The architecture developed by the inventors relies on the use of common photomask sets (or at least one common photomask) to fabricate multiple photonic modules in a single wafer. Photonic modules in multiple wafers can be linked together into a communication platform using optical or electronic means.


