Universal Address System for Distributed Computing Bottlenecks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current processing systems face bottlenecks in computation speed, memory access speed, and storage access, which existing solutions have not adequately addressed, particularly in multiprocessor environments.
Innovation Solution
A distributed computing system with a universal address system and method that utilizes a network of interconnected computing units, each with multiple processing cores, cache memory, and a system memory management unit (SMMU) to manage local and global memory, enabling efficient data retrieval and storage across multiple nodes by mapping virtual addresses to physical and storage addresses based on connectivity and cost parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If memory and storage are centralized in traditional processing systems, then system structure is simplified, but computation speed and memory access speed become bottlenecks
Solution Approach 1:
The patent divides the centralized memory and storage into distributed memory interfaces and storage units across multiple computing nodes. Each node has its own local memory and storage, eliminating the centralized bottleneck and enabling parallel access, which directly improves computation speed without requiring a fundamentally complex new architecture.
Solution Approach 2:
The patent introduces a universal address system that adds a new dimension to memory addressing by mapping virtual addresses to physical addresses across distributed nodes. This dimensional extension allows the system to maintain simplified access semantics while physically distributing memory and storage across multiple nodes, resolving the speed-complexity contradiction.
2Speed
If memory access is centralized through single memory interfaces, then address management is simplified, but memory access speed deteriorates due to bottlenecks
Solution Approach 1:
The patent segments the single centralized memory interface into multiple distributed memory interfaces across different computing nodes. Each node can independently access its local memory without contending for a single interface, thereby improving memory access speed while keeping each individual interface simple.
Solution Approach 2:
The universal address system acts as an intermediary layer between virtual addresses and physical memory locations. It handles the complexity of mapping addresses across distributed nodes transparently, allowing simplified address management at each node while achieving fast distributed memory access.
3Speed
If storage is accessed through traditional disc interfaces, then storage structure is simple, but access speed becomes a bottleneck
Solution Approach 1:
The patent divides centralized disc storage into distributed storage units across multiple nodes. Each node has local storage capacity that can be accessed independently, eliminating the single-point bottleneck of traditional disc interfaces and enabling parallel storage operations that improve access speed.
Solution Approach 2:
The universal address system provides multi-functionality by enabling the same address translation mechanism to work for both memory and storage access. This allows the system to achieve fast distributed storage access without requiring separate complex addressing schemes, maintaining simplicity while improving performance.
4Productivity
If distributed computing nodes are interconnected with universal address mapping, then data access efficiency improves, but address translation complexity increases
Solution Approach 1:
The patent adds a universal address dimension that encompasses both local and remote memory/storage addresses. This dimensional extension allows the system to maintain simple local address translation while providing efficient access to distributed resources through the extended address space, resolving the efficiency-complexity contradiction.
Solution Approach 2:
The universal address system serves as an intermediary that abstracts the complexity of distributed address translation. It provides a unified addressing interface that automatically handles the complexity of mapping to physical locations across nodes, allowing efficient data access without exposing translation complexity to users or applications.
Data Source
AI summary
A distributed computing system that incorporates enhanced distributed storage and a universal address system and method are provided.