Shared Transactional Processing Resource Circuit Area Reduction

Solution Overview

Problem

Transactional memory processing requires additional resources that are costly in terms of silicon real estate and power, especially when the transactional processing functionality is rarely required, leading to significant overhead in circuit area and energy consumption.

Innovation Solution

Sharing transactional processing resources such as data stores, conflict detection circuitry, and instruction decoders among multiple processing elements to limit the number of elements that can process transactions simultaneously, thereby reducing duplication and overhead.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If dedicated transactional processing resources are provided for each processing element, then transaction processing capability is improved, but circuit area and power consumption increase significantly

Engineering Contradiction:

Improvetransaction processing capabilityVSAvoidcircuit area

Core Design Contradiction:

ReliabilityVSArea of stationary object

Solution Approach 1:

Multiple processing elements share a common transactional processing resource (transactional memory unit) instead of each having dedicated resources. This merging approach allows N processing elements to share 1 transactional resource, reducing the total circuit area from O(N) to O(1) for the transactional processing infrastructure while maintaining the ability to service transaction requests from all processing elements.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The shared transactional processing resource is designed to be universal, serving multiple different processing elements through a standardized interface. This multi-functional design allows a single transactional memory unit to handle transactions from any of the N processing elements, eliminating the need for dedicated transactional resources at each processing element and reducing overall circuit area.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Reliability

If dedicated transactional processing resources are provided for each processing element, then transaction processing capability is improved, but power consumption increases

Engineering Contradiction:

Improvetransaction processing capabilityVSAvoidpower consumption

Core Design Contradiction:

ReliabilityVSUse of energy by stationary object

Solution Approach 1:

Multiple processing elements share a common transactional processing resource (transactional memory unit) instead of each having dedicated resources. This merging approach allows N processing elements to share 1 transactional resource, reducing the total circuit area from O(N) to O(1) for the transactional processing infrastructure while maintaining the ability to service transaction requests from all processing elements.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The shared transactional processing resource is designed to be universal, serving multiple different processing elements through a standardized interface. This multi-functional design allows a single transactional memory unit to handle transactions from any of the N processing elements, eliminating the need for dedicated transactional resources at each processing element and reducing overall circuit area.

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Area of stationary object

If transactional processing resources are shared among multiple processing elements, then circuit area and power consumption are reduced, but the number of simultaneous transactions is limited

Engineering Contradiction:

Improvecircuit areaVSAvoidthroughput

Core Design Contradiction:

Area of stationary objectVSProductivity

Solution Approach 1:

The shared transactional processing resource employs dynamic resource allocation and arbitration mechanisms to manage concurrent transaction requests from multiple processing elements. The system dynamically adjusts resource allocation based on demand, using arbitration logic to fairly distribute access to the shared transactional memory unit, thereby maximizing throughput within the constraints of shared resources.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The arbitration mechanism for the shared transactional resource operates in periodic time slots or cycles, allowing multiple processing elements to take turns accessing the shared resource. This periodic scheduling ensures that each processing element gets fair access to the transactional processing capability while maintaining high overall system throughput and utilization.

Inventive Principle:

Principle #19Periodic action

Data Source

PatentUS10908944B2Apparatus with shared transactional processing resource, and data processing method

Publication Date: 2021.02.02 ARM LTD

AI summary

An apparatus (2) with multiple processing elements (4, 6, 8) has shared transactional processing resources (10, 50, 75) for supporting processing of transactions, which comprise operations performed speculatively following a transaction start event whose results are committed following a transaction end event. The transactional processing resources may have a significant overhead and sharing these between the processing elements helps reduce energy consumption and circuit area.