Serial Differential Interconnect for Blocking Link State Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current interconnect architectures in high-performance computing systems face challenges in meeting the increasing demand for bandwidth and power efficiency, particularly in servers and mobile devices, where traditional interconnects struggle to balance performance with power consumption.
Innovation Solution
The development of a High Performance Interconnect (HPI) architecture that employs a layered protocol stack, including a transaction layer, link layer, and physical layer, with features such as credit-based flow control, virtual channels, and a coherence protocol to enhance data transfer efficiency and manage power effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional multi-drop buses are used for electrical communications, then device complexity is reduced, but communication performance and bandwidth are insufficient
Solution Approach 1:
The interconnect architecture is segmented into multiple independent point-to-point links rather than a shared bus. Each link has dedicated bandwidth and can operate independently, eliminating the contention and serialization inherent in multi-drop buses. This segmentation enables parallel data transmission and significantly improves communication performance.
Solution Approach 2:
The architecture transitions from a single shared communication dimension (bus) to multiple parallel communication dimensions (point-to-point links). This dimensional expansion allows simultaneous data transfer across multiple channels, increasing bandwidth and communication throughput without requiring complex arbitration protocols.
2Productivity
If processing power and number of devices are increased, then computing capability is improved, but power consumption increases
Solution Approach 1:
The interconnect employs periodic credit-based flow control mechanisms where receivers signal when their buffers are ready to accept more data. This periodic acknowledgment system enables efficient bandwidth utilization by activating transmission only when needed, reducing idle power consumption while maintaining high computing capability.
Solution Approach 2:
The architecture dynamically adjusts transmission parameters based on system state, including variable data rates and selective activation of communication links. By changing operational parameters rather than maintaining constant high-power transmission, the system achieves high computing capability with reduced average power consumption.
3Speed
If data transfer rate is increased, then communication performance is improved, but power consumption increases
Solution Approach 1:
The interconnect architecture dynamically adjusts data transfer rates based on actual communication needs rather than operating at constant maximum speed. The credit-based flow control mechanism enables the system to reduce transmission rate during low-demand periods while maintaining the capability for high-speed transfer when data buffers require rapid evacuation, thus reducing average power consumption.
Solution Approach 2:
The architecture maintains continuous data flow where possible through overlapping transmissions and efficient buffer management, reducing idle time and power consumption. By keeping the data path actively utilized rather than repeatedly activating and deactivating, the system achieves high effective transfer rates with lower average power consumption.
Data Source
AI summary
A physical layer (PHY) is coupled to a serial, differential link that is to include a number of lanes. The PHY includes a transmitter and a receiver to be coupled to each lane of the number of lanes. The transmitter coupled to each lane is configured to embed a clock with data to be transmitted over the lane, and the PHY periodically issues a blocking link state (BLS) request to cause an agent to enter a BLS to hold off link layer flit transmission for a duration. The PHY utilizes the serial, differential link during the duration for a PHY associated task selected from a group including an in-band reset, an entry into low power state, and an entry into partial width state.