Switching Device Memory Migration Logic for Latency Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In computer systems with multiple processors, remote memory referencing leads to increased latency and reduced performance due to the time-consuming data transfer between nodes, which is not efficiently addressed by existing technologies.

Innovation Solution

A system with a switching device that includes migration logic to identify and migrate frequently accessed memory segments from individual nodes to a central switch memory, allowing processors to access these segments directly from the switch memory, thereby reducing latency and improving performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If remote memory referencing is used to access memory assigned to other nodes, then system capacity and distribution are improved, but latency increases and performance decreases

Engineering Contradiction:
Improvesystem capacityVSAvoidlatency
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system proactively migrates memory segments to the switching device before they are actually needed by processors. The migration logic monitors memory access patterns and anticipates future access needs, performing memory segment migration in advance to ensure that when processors need to access remote memory, the data is already positioned in the switching device, thereby eliminating latency without sacrificing system capacity

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The switching device is enhanced with migration logic that acts as an intermediary between node memories and processors. This intermediary component proactively manages memory segment movement based on predicted access patterns, mediating between the distributed memory architecture (which provides capacity) and the need for fast access (which reduces latency)

Inventive Principle:
Principle #24Intermediary (Mediator)

2Speed

If memory segments are migrated to the switching device, then access speed is improved, but device complexity increases

Engineering Contradiction:
Improveaccess speedVSAvoidswitching device complexity
Core Design Contradiction:
SpeedVSDevice complexity

Solution Approach 1:

The memory system is segmented into movable memory segments that can be independently migrated to the switching device. This segmentation allows the switching device to handle only specific memory segments that are predicted to be frequently accessed, rather than managing entire memory spaces, thereby improving access speed for critical data while limiting the complexity increase to only the necessary switching logic components

Inventive Principle:
Principle #1Segmentation

3Loss of time

If memory segments are migrated to the switching device, then latency is reduced, but the system requires predictive capabilities increasing complexity

Engineering Contradiction:
ImprovelatencyVSAvoidmigration logic complexity
Core Design Contradiction:
Loss of timeVSDevice complexity

Solution Approach 1:

The migration logic performs preliminary analysis of memory access patterns to predict which memory segments will be frequently accessed. By performing this prediction and migration action in advance, the system reduces latency for anticipated accesses while keeping the complexity of the migration logic focused on pattern recognition and timing decisions rather than complex real-time arbitration

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS8078818B2Method and system for migrating memory segments
Publication Date: 2011.12.13 HEWLETT PACKARD ENTERPRISE DEV LP
  • US8078818B2 patent drawing
  • US8078818B2 patent drawing
  • US8078818B2 patent drawing

AI summary

A system comprises a plurality of nodes coupled together via a switching device. Each node comprises a processor coupled to a memory. Migration logic in the switching device is configured to migrate segments of each memory to the switching device.