Remote Memory Throttling by Connectivity Bandwidth

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing computer systems face challenges in seamlessly integrating and managing memory resources across different devices, particularly in scenarios where applications need to access memory beyond their local capacity, requiring complex programming and inefficient use of memory resources.

Innovation Solution

Implementing inter-operating system memory services over communication networks, allowing devices to borrow and lend memory, with operating systems bridging the gap to provide a unified address space and manage memory access without distinction between local and borrowed memory, using techniques like page table manipulation and cache coherency to facilitate seamless byte-access.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If memory resources are expanded across multiple devices, then memory capacity is improved, but system complexity increases

Engineering Contradiction:
Improvememory capacityVSAvoidsystem complexity
Core Design Contradiction:
Quantity of substanceVSDevice complexity

Solution Approach 1:

The patent combines memory resources from multiple devices into a unified memory space that appears as a single contiguous address space to applications. The operating system merges local memory and remote memory accessed via RDMA into one unified memory manager, eliminating the need for applications to distinguish between local and remote memory locations.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces an intermediary operating system layer that manages the complexity of distributed memory access. The OS includes a unified memory manager that handles page tables, translation lookaside buffers, and memory allocation across devices, shielding applications from the underlying complexity while enabling expanded memory capacity.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Ease of operation

If memory access is made transparent across devices, then ease of operation is improved, but bandwidth usage efficiency deteriorates

Engineering Contradiction:
Improvememory access transparencyVSAvoidbandwidth usage efficiency
Core Design Contradiction:
Ease of operationVSLoss of energy

Solution Approach 1:

The patent implements local quality by creating contiguous memory regions that are locally optimized for specific access patterns. The unified memory manager can identify which memory pages are frequently accessed and keep them in local memory or in RDMA memory regions with optimized access paths, while less frequently accessed pages can be stored in remote memory.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent employs dynamic memory management where the system can adaptively move memory pages between local and remote storage based on access patterns. The translation lookaside buffer dynamically updates mappings, and the memory manager can relocate pages to optimize bandwidth usage while maintaining transparent access for applications.

Inventive Principle:
Principle #15Dynamics

3Adaptability or versatility

If contiguous virtual memory regions are mapped to non-contiguous physical memory, then adaptability is improved, but manufacturing precision requirements increase

Engineering Contradiction:
Improvememory mapping flexibilityVSAvoidaddress translation precision
Core Design Contradiction:
Adaptability or versatilityVSManufacturing precision

Solution Approach 1:

The patent segments the virtual address space into manageable pages and uses page tables to map these pages to physical memory locations. The translation lookaside buffer further segments the address translation process by caching recent mappings, allowing the system to handle non-contiguous physical memory while maintaining contiguous virtual memory regions through hierarchical page table structures.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS20260093642A1Throttle Memory as a Service based on Connectivity Bandwidth
Publication Date: 2026.04.02 MICRON TECHNOLOGY INC
  • US20260093642A1 patent drawing
  • US20260093642A1 patent drawing
  • US20260093642A1 patent drawing

AI summary

Systems, methods and apparatuses to throttle network communications for memory as a service are described. For example, a computing device can borrow an amount of random access memory of the lender device over a communication connection between the lender device and the computing device. The computing device can allocate virtual memory to applications running in the computing device, and configure at least a portion of the virtual memory to be hosted on the amount of memory loaned by the lender device to the computing device. The computing device can throttle data communications used by memory regions in accessing the amount of memory over the communication connection according to the criticality levels of the contents stored in the memory regions.