Partial Data Transfers with Chunk Valid Identifiers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current protocols, such as AXI, do not support partial data transfers from a narrower data bus to a wider data bus, leading to systems with mismatched data bus sizes that require additional merge-buffers, resulting in increased area and power requirements.
Innovation Solution
Implementing a system where data beats are tagged with a 'Chunk Valid' identifier to enable partial data transfers on a wider data bus, allowing for the elimination of merge buffers and optimizing area and power metrics by enabling parallel data transfer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If protocols do not provide for transferring partial data from one data bus to a second data bus with increased capacity, then data transfer simplicity is maintained, but additional merge-buffers are required at all points of interconnect, resulting in increased area and power requirements
Solution Approach 1:
The data bus transfer is segmented into partial data transfers using chunk valid identifiers. Each identifier indicates which portions of the wider data bus contain valid data, allowing the transfer to be divided into manageable segments that match the narrower source bus width without requiring merge-buffers.
Solution Approach 2:
The protocol introduces new parameters (chunk valid identifiers with multiple bits) that change the state of data transfer from all-or-nothing to partial-validity indication. This allows the receiving end to correctly interpret which portions of the wider bus contain valid data, enabling partial transfers without additional buffering infrastructure.
2Use of energy by stationary object
If protocols do not provide for transferring partial data from one data bus to a second data bus with increased capacity, then data transfer protocol simplicity is maintained, but power requirements increase due to additional merge-buffers
Solution Approach 1:
By segmenting the data transfer into partial transfers with validity indicators, the system eliminates the need for power-consuming merge-buffers. The chunk valid identifiers allow the receiving end to process only the valid portions of data without requiring additional buffering hardware that would consume power.
Solution Approach 2:
The harmful element (merge-buffers) is extracted from the system by implementing partial data transfer capability directly in the protocol. The chunk valid identifiers enable the protocol itself to handle the complexity of mismatched bus widths without requiring additional physical components.
3Adaptability or versatility
If merge-buffers are added at all points of interconnect to handle data bus size mismatches, then data transfer compatibility is improved, but system area increases
Solution Approach 1:
The chunk valid identifier mechanism provides a universal solution for handling data bus size mismatches of any width combination. Rather than requiring specific merge-buffer hardware for each mismatch scenario, the protocol-level identifier system universally handles all cases of partial data transfers between buses of different widths.
Solution Approach 2:
The chunk valid identifiers act as an intermediary mechanism between the narrower source bus and wider destination bus. Instead of using physical merge-buffers as intermediaries, the protocol uses these identifiers to mediate the transfer, indicating which portions of the wider bus contain valid data from the narrower source.
4Adaptability or versatility
If merge-buffers are added at all points of interconnect to handle data bus size mismatches, then data transfer compatibility is improved, but power requirements increase
Solution Approach 1:
The power-consuming merge-buffers are extracted from the system architecture and replaced with a protocol-level solution using chunk valid identifiers. This extraction eliminates the continuous power consumption associated with maintaining buffer memory and control logic at all interconnect points while preserving data bus compatibility.
Solution Approach 2:
The physical hardware system (merge-buffers as mechanical/electrical components) is replaced with a protocol-level software/logic system (chunk valid identifiers). This substitution transitions from a hardware-based solution that consumes power to a protocol-based solution that achieves the same compatibility function with minimal additional power consumption.
Data Source
AI summary
A system, apparatus and method for an interface based system that may be composed of a diverse set of blocks with different data bus sizes. These different data bus sizes can be optimized by permitting partial data transfers on the different sized buses.