Columnar PLD Interface Routing for Bandwidth Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current programmable logic devices (PLDs) like FPGAs face limitations in efficiently managing interconnects and signal routing due to the need for long-distance signal transmission across the device, which can lead to bandwidth constraints and resource inefficiencies.
Innovation Solution
The implementation of multiple columnar interfaces on a substrate, coupled with an interposer, allows for parallel signal routing and reduced bandwidth requirements by distributing interconnects and control circuits, enabling more efficient signal management and resource conservation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If signals are transmitted across the entire PLD device, then all logic elements can be accessed, but bandwidth constraints and resource inefficiencies occur due to long transmission distances
Solution Approach 1:
The patent divides the PLD interface into multiple columnar interfaces (first columnar interface, second columnar interface, etc.) that are spaced apart and extend parallel to each other through the substrate. Each columnar interface serves a specific region of logic elements, segmenting the monolithic interface into distributed access points. This segmentation reduces transmission distances and enables parallel signal routing, resolving the contradiction between universal access and bandwidth efficiency.
Solution Approach 2:
The patent transitions from a planar interface arrangement to a three-dimensional columnar structure that extends through the substrate depth. The columnar interfaces are positioned at different locations and heights, creating vertical routing paths that bypass the limitations of surface-level signal transmission. This dimensional change enables shorter signal paths and parallel routing capabilities, improving bandwidth efficiency while maintaining access to all logic elements.
2Productivity
If multiple columnar interfaces are implemented, then signal routing efficiency improves, but device complexity increases
Solution Approach 1:
Each columnar interface is designed with identical functional capabilities, including interconnects, control circuits, driver circuits, rebuffering circuits, signal conditioning circuits, deskewing circuits, clock synchronization circuits, power management circuits, and testing/debugging circuits. This universal design allows any columnar interface to handle any signal type, reducing the need for specialized routing logic and simplifying the overall control architecture despite the increased number of interfaces.
Solution Approach 2:
The patent uses replicated columnar interface structures that are spaced apart and extend parallel to each other. Each columnar interface is essentially a copy of the others, containing the same circuit components and functionality. This copying approach standardizes the interface design, reduces design complexity through reuse, and enables parallel signal processing while maintaining consistent performance across all interfaces.
Data Source
AI summary
A PLD comprises a substrate, an array of programmable logic elements formed in the substrate, a first columnar interface coupling to the array of logic elements and extending in the substrate substantially parallel to a first side of the substrate, and at least a second columnar interface coupling to the array of logic elements and extending in the substrate substantially parallel to the first columnar interface. The interfaces illustratively provide a plurality of interconnects, control circuits and one or more of driver circuits, rebuffering circuits, signal conditioning circuits, deskewing circuits, clock synchronization circuits, power management circuits, testing/debugging circuits, partial reconfiguration circuits, multi-plexing circuits, pipelining circuits and storage circuits. The PLD is mounted on an interposer so that its interfaces electrically couple to electrically conducting paths on the interposer.


