Interconnect Circuitry for Multi-Transaction Routing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Modern data processing systems face inefficiencies in memory bandwidth utilization due to cache coherency issues and the need for multiple smaller burst transactions, which increase power consumption and reduce memory efficiency.

Innovation Solution

A data processing system that allows a master device to issue multi-transaction requests, enabling multiple transactions to be routed to appropriate slave devices for processing, with each slave device providing independent responses, thereby optimizing resource utilization and bandwidth efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If multiple smaller burst transactions are issued to maintain cache coherency, then cache coherency is maintained, but memory bandwidth utilization decreases and power consumption increases

Engineering Contradiction:
Improvecache coherencyVSAvoidmemory bandwidth utilization
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The system segments the multi-transaction request into individual transaction requests for each slave device, allowing parallel processing while maintaining coherency. The interconnect circuitry divides the address range into portions and routes them to appropriate slave devices independently.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system merges multiple individual transaction responses into a single multi-transaction response returned to the master device. This combines the results from multiple slave devices into one unified response, reducing the number of response handling operations.

Inventive Principle:
Principle #5Merging (Combining)

2Reliability

If multiple smaller burst transactions are issued to maintain cache coherency, then cache coherency is maintained, but power consumption increases

Engineering Contradiction:
Improvecache coherencyVSAvoidpower consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system segments the multi-transaction request into individual transaction requests for each slave device, allowing parallel processing while maintaining coherency. The interconnect circuitry divides the address range into portions and routes them to appropriate slave devices independently.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system enables continuous processing of multiple transactions by different slave devices in parallel, eliminating idle time between transactions. This keeps the memory subsystem continuously utilized with larger effective burst sizes.

Inventive Principle:
Principle #20Continuity of useful action

3Device complexity

If transactions are limited to single cache line size, then cache coherency is simplified, but the number of transactions required increases

Engineering Contradiction:
Improvecache coherency complexityVSAvoidnumber of transactions
Core Design Contradiction:
Device complexityVSProductivity

Solution Approach 1:

The interconnect circuitry is designed to handle multiple functions: it distributes address portions to appropriate slave devices, tracks responses from multiple slave devices, and reassembles them into a unified multi-transaction response. This single component performs what would otherwise require multiple separate mechanisms.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The interconnect circuitry acts as an intermediary between the master device and multiple slave devices. It receives the multi-transaction request, coordinates the distribution to slave devices, collects responses, and returns the consolidated result, simplifying the interface for the master device.

Inventive Principle:
Principle #24Intermediary (Mediator)

4Reliability

If multiple smaller burst transactions are used, then cache coherency is maintained, but re-ordering capabilities of memory controllers are reduced

Engineering Contradiction:
Improvecache coherencyVSAvoidre-ordering capabilities
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The system segments the multi-transaction request into individual transaction requests for each slave device, allowing parallel processing while maintaining coherency. The interconnect circuitry divides the address range into portions and routes them to appropriate slave devices independently.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system enables dynamic re-ordering of transactions by different slave devices based on their individual processing capabilities and queue states. Each slave device can independently reorder its assigned transactions while the overall multi-transaction operation maintains coherence.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3095037B1A data processing system and method for handling multiple transactions
Publication Date: 2018.08.08 ARM LTD
  • EP3095037B1 patent drawingFigure 1A~1B
  • EP3095037B1 patent drawingFigure 2
  • EP3095037B1 patent drawingFigure 3

AI summary

A data processing system having a master device (55) and a plurality of slave devices (60,62,64) uses interconnect circuitry (50) to couple the master device with the plurality of slave devices to enable transactions to be performed by the slave devices upon request from the master device. The master device issues a multi-transaction request identifying multiple transactions to be performed, the multi- transaction request providing a base transaction identifier, a quantity indication indicating a number of transactions to be performed, and address information. Request distribution circuitry within the interconnect circuitry analyses the address information and the quantity indication in order to determine, for each of the multiple transactions, the slave device that is required to perform that transaction.