Remote Memory Allocation via Multi-Path Fabric Node Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multi-processing systems, existing memory allocation technologies face challenges in efficiently managing virtual memory across interconnected nodes, particularly in selecting the optimal node for remote memory allocation based on performance indicators like latency, availability, and fabric congestion, which affects system performance and resource utilization.
Innovation Solution
A memory allocation device interconnected via a load-store type multi-path fabric dynamically allocates both local and remote memory pages, selecting the optimal remote node based on performance indicators such as latency, availability, bandwidth, data access patterns, and fabric congestion to optimize memory allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If memory allocation device selects remote node based on multiple performance indicators, then memory allocation efficiency is improved, but device complexity increases
Solution Approach 1:
The patent segments the memory allocation function by separating local memory allocation (handled by local memory allocation device) from remote memory allocation (handled by remote memory allocation device). This segmentation allows each device to specialize in its respective function, improving overall memory allocation efficiency while distributing system complexity across multiple components rather than concentrating it in a single device.
Solution Approach 2:
The fabric interconnected processing system acts as an intermediary between local and remote memory allocation devices. This intermediary fabric manages the communication and data transfer between nodes, enabling efficient remote memory allocation while abstracting the complexity of inter-node communication from the memory allocation logic itself.
2Productivity
If memory allocation device uses multiple performance indicators for node selection, then resource utilization is improved, but measurement precision requirements increase
Solution Approach 1:
The patent implements feedback mechanisms where memory allocation devices monitor performance indicators (such as latency, bandwidth, and availability) of remote nodes and use this feedback information to make dynamic allocation decisions. This feedback loop enables the system to adapt to changing conditions and optimize resource utilization while using practical, measurable performance metrics rather than requiring extremely precise measurements.
Solution Approach 2:
The system dynamically changes allocation parameters based on observed performance indicators. Instead of requiring fixed, highly precise measurements, the system adjusts allocation decisions based on varying performance parameters such as current latency, available bandwidth, and node availability, allowing flexible resource utilization with practical measurement capabilities.
Data Source
AI summary
A multi-path fabric interconnected system with many nodes and many communication paths from a given source node to a given destination node. A memory allocation device on an originating node (local node) requests an allocation of memory from a remote node (i.e., requests a remote allocation). The memory allocation device on the local node selects the remote node based on one or more performance indicators. The local memory allocation device may select the remote node to provide a remote allocation of memory based on one or more of: latency, availability, multi-path bandwidth, data access patterns (both local and remote), fabric congestion, allowed bandwidth limits, maximum latency limits, and, available memory on remote node.