Multi-byte Bus-invert Coding for SSO Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional bus-invert coding methods are inadequate in reducing simultaneous switching output (SSO) noise, which occurs when multiple bus drivers switch their outputs simultaneously, leading to significant power supply noise in high-bandwidth memory systems.
Innovation Solution
The proposed technique utilizes expanded encoding spaces for multi-byte data transmissions by determining Hamming Distance across multiple bytes and associated DBI bits, reducing SSO noise by selectively inverting data bits to minimize aggregate transitions, thereby reducing SSO noise by up to 87.5% as bus width doubles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If conventional bus-invert coding is used to reduce SSO noise, then noise reduction is achieved by inverting data bits when more than half the bits switch polarity, but the noise reduction is limited to 50% and does not scale effectively with increased bus width
Solution Approach 1:
The patent merges multiple independent byte encoding operations into a unified multi-byte encoding process. By treating multiple bytes as a single encoded unit and computing Hamming Distance across the entire multi-byte data set, the system achieves coordinated bit transitions that reduce aggregate switching activity. This combining approach allows the encoder to optimize transitions across byte boundaries, achieving up to 87.5% noise reduction on 32-bit buses compared to the limited 50% reduction of conventional per-byte encoding.
Solution Approach 2:
The patent extends the encoding dimension from individual bytes to multi-byte data sets. By computing Hamming Distance across multiple bytes simultaneously and using multi-byte look-up tables, the system adds a dimensional aspect to the encoding process that enables broader optimization of bit transitions. This dimensional expansion allows the encoder to consider transitions across the entire data width rather than being constrained to individual byte boundaries.
2Object-affected harmful factors
If multi-byte encoding is implemented to reduce SSO noise further, then noise reduction increases to 87.5% for 32-bit buses, but the encoding complexity and computational overhead increase
Solution Approach 1:
The patent pre-computes Hamming Distance values and stores them in multi-byte look-up tables during system initialization or manufacturing. These pre-computed tables contain optimal encoding decisions for various multi-byte data patterns, allowing the runtime encoder to simply perform table lookups rather than performing complex Hamming Distance calculations for each data transmission. This preliminary action significantly reduces the computational overhead of multi-byte encoding while maintaining the noise reduction benefits.
3Object-affected harmful factors
If Hamming Distance is computed across multiple bytes to minimize aggregate transitions, then SSO noise reduction is maximized, but the computational resources and processing time required increase
Solution Approach 1:
The patent pre-computes and stores Hamming Distance values in multi-byte look-up tables, transforming complex runtime calculations into simple table lookup operations. This allows the encoder to determine optimal bit transitions for multi-byte data without performing real-time Hamming Distance computations, significantly reducing processing time while maintaining optimal noise reduction performance.
Solution Approach 2:
The patent replaces the mechanical/computational process of real-time Hamming Distance calculation with a lookup-based system. Instead of dynamically computing bit transition counts across multiple bytes, the system substitutes this computational mechanism with pre-computed reference data stored in tables, dramatically reducing the processing time and computational resources required for encoding decisions.
Data Source
AI summary
An encoding process for bus data utilizes data from multiple data line groups on a multi-byte wide bus where each group has an associated DBI line. The process leverages the expanded encoding space for the multiple groups and associated multiple DBI bits. This process may be expanded to four bytes, eight bytes, etc.