Interrupted Instruction Migration Using Model Metadata

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In computing environments, interruptible instructions that require a significant number of execution cycles to complete often lead to inefficiencies, as they necessitate additional processing and metadata recreation upon resumption, which can be time-consuming and resource-intensive.

Innovation Solution

A computer program product that determines if model-dependent metadata is needed to re-execute an interrupted instruction, using it if the processor models have a predefined relationship, thereby reducing storage accesses and improving performance by avoiding redundant metadata generation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If an interruptible instruction is resumed on a different processor, then processing continuity is maintained, but metadata recreation is required which reduces performance

Engineering Contradiction:
Improveinstruction completion rateVSAvoidmetadata recreation time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent stores model-dependent metadata in a parameter block during the first processor's execution of the interruptible instruction. This preliminary action ensures that when the instruction is resumed on a different processor model, the metadata is already available and does not need to be recreated, thus resolving the contradiction by preparing the metadata in advance

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent copies the model-dependent metadata from the first processor's parameter block to the second processor's parameter block. This copying mechanism allows the resuming processor to utilize existing metadata instead of recreating it, maintaining processing continuity while avoiding the time loss associated with metadata recreation

Inventive Principle:
Principle #26Copying

2Productivity

If model-dependent metadata is stored and reused, then storage accesses are reduced, but compatibility checks between processor models are required

Engineering Contradiction:
Improveprocessing speedVSAvoidprocessor compatibility verification
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent performs a model version comparison check before reusing the stored metadata. This preliminary compatibility verification ensures that the metadata from the first processor can be safely used by the second processor, resolving the contradiction by establishing compatibility criteria in advance before metadata reuse

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces a model version number as an intermediary element that mediates between different processor models. By comparing model version numbers, the system determines whether metadata can be reused without full compatibility verification, thus reducing device complexity while maintaining processing speed

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3877842B1Migration of partially completed instructions
Publication Date: 2023.01.11 INTERNATIONAL BUSINESS MACHINE CORPORATION
  • EP3877842B1 patent drawingFigure 1A~1B
  • EP3877842B1 patent drawingFigure 2
  • EP3877842B1 patent drawingFigure 3A~3F

AI summary

Migration of partially completed instructions. A determination is made that processing of an operation of an instruction executing on a processor has been interrupted prior to completion. The instruction is re-executed on a selected processor to resume forward processing of the instruction. The re-executing includes determining whether model-dependent metadata is to be used by the selected processor in re-executing the instruction. Based on determining the model-dependent metadata is to be used, the model-dependent metadata is used in re-executing the instruction. Based on determining the model-dependent metadata is not to be used, proceeding with re-executing the instruction without using the model-dependent metadata.