NoC Switch Dynamic Width Adaptation for Variable Data Flits
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Network-on-Chip (NoC) architectures are limited in handling data of different sizes, requiring fixed data widths and necessitating width conversion or bifurcation of the NoC, which can lead to deadlock conditions.
Innovation Solution
The proposed NoC architecture includes a plurality of NoC master and slave circuits interconnected by switches that can dynamically adjust to different data widths by implementing specific operating modes, allowing concurrent handling of flits of various sizes without the need for width conversion or bifurcation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional NoC architectures use fixed data widths, then the system structure is simple and stable, but the adaptability to handle different data sizes is poor
Solution Approach 1:
The NoC switch is designed to dynamically adjust its data width based on the incoming data requirements. The switch can operate in different modes (e.g., 128-bit mode, 256-bit mode) and automatically selects the appropriate mode based on the data width detected from master circuits, eliminating the need for fixed width configurations or bifurcation schemes.
Solution Approach 2:
The switch is designed as a universal component that can handle multiple data widths (128-bit, 256-bit, and other sizes) within a single unified architecture. This multi-functional design allows the same switch infrastructure to serve diverse data transmission requirements without requiring separate specialized paths or conversion circuits.
2Adaptability or versatility
If width conversion or bifurcation is implemented to handle different data sizes, then adaptability improves, but deadlock conditions may occur and system reliability deteriorates
Solution Approach 1:
The switch dynamically configures its internal data paths and buffering structures based on the detected data width. When a 256-bit data unit is detected, the switch activates appropriate wide paths and buffers; when 128-bit units are detected, it uses the corresponding narrower paths. This dynamic adaptation eliminates the need for static width conversion or bifurcation that could lead to deadlock conditions.
3Productivity
If fixed data width paths are used, then the system is simpler to implement, but bandwidth utilization is reduced when handling variable size data
Solution Approach 1:
The switch internally reconfigures its data paths, buffering capacity, and transmission timing based on the detected data width. For 256-bit data, the switch activates wider paths and adjusts timing accordingly; for 128-bit data, it uses appropriate narrower paths. This dynamic reconfiguration allows full bandwidth utilization for each data type without requiring complex external width conversion or bifurcation infrastructure.
Data Source
AI summary
A network-on-chip (NoC) includes a switch. The switch includes a first sub-switch, a second sub-switch, and a synchronization channel coupled to the first sub-switch and the second sub-switch. The first sub-switch and the second sub-switch are coupled to corresponding sub-switches in at least one other switch included in the NoC. Each of the first sub-switch and the second sub-switch includes ports in north, south, east, and west directions. The first sub-switch and the second sub-switch exchange flits of data through an additional port of the first sub-switch coupled to an additional port of the second sub-switch.