Mixed-Radix Switch Matrix Architecture for FPGA Interconnect Area

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current integrated circuits, such as FPGAs, face challenges in achieving efficient and flexible interconnect architectures that balance operating speed, die area, and power consumption, particularly in mobile computing applications where 'dark silicon' regions are prevalent due to inflexible dedicated hardware.

Innovation Solution

The integrated circuit employs a mixed-mode interconnect architecture with switch matrices organized in hierarchical and mesh networks, utilizing different radices like radix-4, boundary-less radix-3, and radix-2 configurations to optimize interconnect paths and reduce resource requirements, allowing for efficient communication between computing elements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Area of stationary object

If a traditional uniform interconnect architecture is used, then routing simplicity is maintained, but interconnect area and power consumption increase

Engineering Contradiction:
Improveinterconnect areaVSAvoidinterconnect architecture complexity
Core Design Contradiction:
Area of stationary objectVSDevice complexity

Solution Approach 1:

The interconnect architecture is segmented into multiple stages (first stage, second stage, third stage) with different radix configurations. Each stage handles specific routing requirements, allowing the system to optimize for both area and complexity by dividing the interconnect function into specialized segments rather than using a uniform architecture throughout.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different radix configurations are applied to different stages of the interconnect architecture based on local routing requirements. The first stage uses radix-4 for efficient local connectivity, the second stage uses boundary-less radix-3 for intermediate routing, and the third stage uses radix-2 for extended reach, optimizing each local segment for its specific function.

Inventive Principle:
Principle #3Local quality

2Adaptability or versatility

If dedicated hardware is used, then operating speed is improved, but adaptability decreases leading to dark silicon regions

Engineering Contradiction:
Improvereconfiguration flexibilityVSAvoidpower consumption
Core Design Contradiction:
Adaptability or versatilityVSUse of energy by stationary object

Solution Approach 1:

The interconnect architecture implements dynamic reconfiguration capabilities where the routing paths and radix configurations can be changed during operation. This allows the system to adapt to different computational workloads and activate only the necessary interconnect resources, preventing dark silicon regions by ensuring that hardware is actively used when needed and can be reconfigured for new functions.

Inventive Principle:
Principle #15Dynamics

3Speed

If higher radix configurations are used, then interconnect speed is improved, but resource requirements increase

Engineering Contradiction:
Improveinterconnect speedVSAvoidinterconnect resources
Core Design Contradiction:
SpeedVSQuantity of substance

Solution Approach 1:

The architecture changes the radix parameter across different stages to optimize the balance between speed and resources. Higher radix (radix-4) is used in the first stage where speed is critical for local connectivity, while lower radix (radix-2) is used in the third stage where resource efficiency is more important for extended routing, creating an optimized gradient of radix values.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3259843B1Mixed-radix and/or mixed-mode switch matrix architecture and integrated circuit, and method of operating same
Publication Date: 2022.03.02 FLEX LOGIX TECHNOLOGIES INC
  • EP3259843B1 patent drawingFigure 1A
  • EP3259843B1 patent drawingFigure 1B
  • EP3259843B1 patent drawingFigure 1C

AI summary

An integrated circuit comprising a plurality of logic tiles, wherein each logic tile includes a plurality of (i) computing elements and (ii) switch matrices. The plurality of switch matrices are arranged in stages including (i) a first stage, configured in a hierarchical network (for example, a radix-4 network), wherein, each switch matrix of the first stage is connected to at least one associated computing element, (ii) a second stage configured in a hierarchical network (for example, a radix-2 or radix-3 network) and coupled to switches of the first stage, and (iii) a third stage configured in a mesh network and coupled to switches of the first and/or second stages. In one embodiment, the third stage of switch matrices is located between the first stage and second stage of switch matrices; in another embodiment, the third stage is the highest stage.