Data Identity Caching with Node-Specific Read-Only Caches
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cache coherency issues in networks of computational nodes, particularly in machine intelligence systems like artificial neural networks, arise due to multiple nodes having different values for the same data element, leading to inconsistencies and inefficiencies.
Innovation Solution
Implementing caches that are uniquely associated with computational nodes based on data identity, ensuring only read-only data is stored in these caches, and using a shared address space with distinct write access for each node to eliminate cache coherency problems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple computational nodes share a common back store and data, then data sharing and collaboration are improved, but cache coherency issues arise leading to inconsistencies
Solution Approach 1:
The patent segments the cache system into node-specific caches, where each computational node has its own dedicated cache memory. This segmentation eliminates cache coherency issues by preventing multiple nodes from having conflicting copies of the same data, while still allowing data sharing through controlled access to the back store.
Solution Approach 2:
The patent implements local quality by giving each computational node its own specialized cache with optimized characteristics for that node's specific computational needs. Each node's cache is tailored to its local requirements, improving performance while maintaining system-wide data consistency through the shared back store.
2Speed
If caches are implemented to speed up data access, then computational speed is improved, but cache coherency maintenance increases system complexity
Solution Approach 1:
The patent extracts the cache coherency management problem by removing shared cache access from the system. Instead of having multiple nodes share and synchronize cache contents, each node has its own private cache, eliminating the need for complex coherency protocols while maintaining fast data access through local caching.
3Reliability
If read-only data is stored in node-specific caches, then data integrity is improved, but cache utilization efficiency may be reduced
Solution Approach 1:
The patent applies preliminary action by pre-loading read-only data into node-specific caches before computational nodes need to access it. This ensures data integrity is maintained while improving cache utilization efficiency, as the cache is proactively populated with data that will be needed, reducing misses and improving performance.
Data Source
AI summary
Systems and methods related to data identity based caching in a network of computational nodes are disclosed herein. A system may include a set of computational nodes, an interconnect fabric that networks the set of computational nodes, and a set of caches. The caches may be uniquely associated with the computational nodes and may store data in the caches based on an identity of the data. For example, the caches may store data based on the data being read-only data, based on the data being in an address range, or based on the data being fixed for a remainder of a complex computation. Basing the storage of data in the caches on the identity of the data may reduce cache coherency issues.