OoO Processor Mapper Snapshots for Low-Power Instruction Unwinding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Out-of-order (OoO) processors face inefficiencies in power consumption and unwinding operations due to frequent access and writing to snapshot circuitry during instruction unwinding, particularly when floating-point architectural registers are used as destinations, which is common in typical programs.

Innovation Solution

The system employs a mapper that intermittently blocks access to floating-point snapshot circuitry based on the presence of floating-point architectural registers used as destinations, using integer and floating-point snapshot circuitry to store and restore mapper states efficiently, thereby reducing unnecessary power consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the processor accesses and writes to floating-point snapshot circuitry during every instruction unwinding operation, then the mapper state restoration is complete and accurate, but power consumption increases due to unnecessary access when floating-point architectural registers are not used as destinations

Engineering Contradiction:
Improvemapper state restoration accuracyVSAvoidpower consumption during unwinding operations
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent applies partial action by selectively accessing floating-point snapshot circuitry only when floating-point architectural registers are used as destinations in unwound instructions. The mapper tracks FP register usage and blocks unnecessary access to FP snapshot circuitry, performing the restoration operation partially (only when needed) rather than universally for all unwinding operations, thereby reducing power consumption while maintaining restoration accuracy when required

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The mapper performs self-service by autonomously tracking which instructions use floating-point architectural registers as destinations and making decisions about whether to access the FP snapshot circuitry. This self-determination mechanism eliminates the need for external control signals and enables the system to automatically optimize power consumption based on actual register usage patterns

Inventive Principle:
Principle #25Self-service

2Reliability

If the processor implements frequent snapshot storage and restoration operations, then instruction unwinding accuracy is maintained, but the complexity of the snapshot circuitry and mapping infrastructure increases

Engineering Contradiction:
Improveinstruction unwinding accuracyVSAvoidsnapshot circuitry complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the snapshot circuitry into separate integer snapshot circuitry and floating-point snapshot circuitry. This segmentation allows independent management and access control for each type, reducing the complexity of the unified system while maintaining the ability to restore both integer and floating-point mapper states accurately during unwinding operations

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS12386625B2System and method for instruction unwinding in an out-of-order processor
Publication Date: 2025.08.12 MARVELL ASIA PTE LTD
  • US12386625B2 patent drawing
  • US12386625B2 patent drawing
  • US12386625B2 patent drawing

AI summary

A system and corresponding method unwind instructions in an out-of-order (OoO) processor. The system comprises a mapper. In response to a restart event causing at least one instruction to be unwound, the mapper restores a present integer mapper state and present floating-point (FP) mapper state, used for mapping instructions, to a former integer mapper state and former FP mapper state, respectively. The mapper stores integer snapshots and FP snapshots of the present integer and FP mapper state, respectively, to expedite restoration to the former integer and FP mapper state, respectively. Access to the FP snapshots is blocked, intermittently, as a function of at least one FP present indicator used by the mapper to record presence of FP registers used as destinations in the instructions. Blocking the access, intermittently, improves power efficiency of the OoO processor.