Serial Interface Line Speed Packet Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

High-speed serial communication systems, such as PCI Express, face latency issues due to the processing time required for header decoding and data management in finite state machine controllers, which can lead to reduced throughput and bandwidth.

Innovation Solution

Implementing a serial interface with independent pipelined logical modules that include header processing, command analysis, and data direction modules to process and transmit transaction packets at line speed, allowing for simultaneous execution of operations based on available header information and data, thereby reducing latency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If a single finite state machine controller is used to process transaction packets, then the interface structure is simple, but the processing speed is limited and latency increases

Engineering Contradiction:
Improvetransaction packet processing speedVSAvoidinterface controller structure
Core Design Contradiction:
SpeedVSDevice complexity

Solution Approach 1:

The patent divides the transaction packet processing into multiple independent pipeline stages: header receipt, data receipt, header decoding, and data management. Each stage operates independently and simultaneously, allowing the interface to process multiple packets at different stages of the pipeline at the same time, thereby achieving line speed processing without increasing overall interface complexity

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements preliminary actions by receiving and buffering header information and data separately before processing. The header is received and decoded in advance while data is simultaneously received and buffered, allowing both operations to proceed in parallel and eliminating sequential processing delays

Inventive Principle:
Principle #10Preliminary action

2Productivity

If header decoding and data management are performed sequentially, then the processing logic is simple, but throughput is reduced due to latency

Engineering Contradiction:
Improvesystem throughputVSAvoidprocessing latency
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent maintains continuous useful action by implementing a pipeline architecture where header decoding and data management operations occur simultaneously and continuously. While one packet's header is being decoded, another packet's data can be received and buffered, ensuring that the interface is always productive and eliminating idle time between operations

Inventive Principle:
Principle #20Continuity of useful action

Solution Approach 2:

The patent performs preliminary actions by receiving and buffering the header information before the complete packet arrives, and simultaneously buffering data as it streams in. This allows the decoding and management operations to begin in advance and proceed in parallel, significantly reducing the total processing time for each transaction packet

Inventive Principle:
Principle #10Preliminary action

3Speed

If the interface cannot process packets at line speed, then idle characters must be inserted to manage latency, but this reduces maximum bandwidth

Engineering Contradiction:
Improveline speed processing capabilityVSAvoidprocessing architecture
Core Design Contradiction:
SpeedVSDevice complexity

Solution Approach 1:

The patent segments the transaction packet into header and data portions that are received and processed through separate pipeline stages. The header receipt stage and data receipt stage operate independently and simultaneously, allowing the interface to maintain line speed processing without requiring idle characters to manage latency

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent adds a temporal dimension to the processing architecture by implementing a multi-stage pipeline that processes different packets at different times simultaneously. This allows the interface to handle multiple transactions in parallel across different pipeline stages, achieving line speed processing capability without increasing the complexity of the base interface architecture

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentUS8984193B1Line speed sequential transaction packet processing
Publication Date: 2015.03.17 UNISYS CORP
  • US8984193B1 patent drawing
  • US8984193B1 patent drawing
  • US8984193B1 patent drawing

AI summary

Methods and systems for processing transaction packets at a serial interface are disclosed. The method includes receiving transaction information at a serial interface. The method further includes executing one or more pipelined operations based on the transaction information, where the operations relate to processing of the transaction packet. The method is performed such that the serial interface is configured to send and receive transaction packets at a line speed of the serial bus.