Common PHY Architecture for Multi-Protocol Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As data rates for serial links exceed 32.0 GT/s, existing interconnect architectures face challenges in efficiently managing error correction and latency variations across different protocols like PCIe, CXL, and UPI, leading to inefficiencies in energy consumption and performance.
Innovation Solution
A common physical layer (PHY) is introduced that supports multiple interconnect protocols, dynamically adjusting Forward Error Correction (FEC) and Cyclic Redundancy Check (CRC) mechanisms based on operating conditions to optimize error correction and latency according to specific protocol needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple separate PHYs are used for different interconnect protocols (PCIe, CXL, UPI), then each protocol can have optimized error correction and latency characteristics, but device complexity and energy consumption increase due to redundant error correction mechanisms
Solution Approach 1:
The patent implements a universal PHY architecture that can operate with multiple interconnect protocols (PCIe, CXL, UPI) simultaneously. The common PHY includes configurable error correction mechanisms that can be dynamically adjusted based on the active protocol, eliminating the need for separate dedicated PHY instances for each protocol while maintaining protocol-specific optimization
Solution Approach 2:
The patent merges previously separate PHY instances for different protocols into a single common PHY. The error correction functionality is consolidated into shared hardware resources that can be dynamically allocated and configured based on protocol requirements, reducing overall device complexity while preserving protocol-specific performance characteristics
2Loss of time
If protocol-specific PHYs are used, then latency can be optimized for each protocol, but energy consumption increases due to redundant error correction mechanisms operating simultaneously
Solution Approach 1:
The patent implements dynamic configuration of error correction mechanisms within the common PHY. The error correction strength, latency parameters, and operational modes can be dynamically adjusted based on which protocol is currently active, allowing the system to optimize for low latency when needed while reducing error correction overhead to save energy when protocol-specific optimizations are less critical
Solution Approach 2:
The patent changes operational parameters of the error correction mechanisms based on the active protocol. Different protocols can have different error correction codes, block sizes, and retransmission parameters applied, allowing each protocol to operate with optimal latency characteristics while sharing the same physical layer hardware
3Reliability
If redundant error correction mechanisms are implemented for each protocol, then reliability is improved, but device complexity and energy consumption increase
Solution Approach 1:
The common PHY includes intelligent protocol detection and automatic configuration capabilities. When a specific protocol is detected, the system automatically configures the appropriate error correction parameters and activates only the necessary correction mechanisms for that protocol, eliminating the need for all error correction mechanisms to operate simultaneously and reducing energy waste
Data Source
AI summary
Systems and devices can include a physical layer (PHY) that includes a logical PHY to support multiple interconnect protocols. The logical PHY can include a first set of cyclic redundancy check (CRC) encoders corresponding to a first interconnect protocol, and a second set of CRC encoders corresponding to a second interconnect protocol. A multiplexer can direct data to the first set or the second set of CRC encoders based on a selected interconnect protocol. The logical PHY can include a first set of error correcting code (ECC) encoders corresponding to the first interconnect protocol and a second set of ECC encoders corresponding to the second interconnect protocol. The multiplexer can direct data to the first set or the second set of ECC encoders based on the selected interconnect protocol. In embodiments, different CRC/ECC combinations can be used based on the interconnect protocol and the link operational conditions.


