Column-Read Memory for Sparse Hashing Similarity Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large-scale similarity searches in machine learning and artificial intelligence are compute and memory intensive due to the sheer volume and richness of data, making existing techniques inefficient.
Innovation Solution
A compute device equipped with column-read enabled memory and media access circuitry that performs stochastic associative searches by accessing individual bits of data in a three-dimensional cross-point architecture, enabling efficient similarity searches with minimal bus usage and power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional memory architectures are used for large-scale similarity searches, then data storage capacity is sufficient, but computational load and power consumption become excessive
Solution Approach 1:
The patent introduces column-read enabled memory as an intermediary between traditional memory and processing units. This memory architecture enables direct columnar data access patterns that eliminate the need for complex data movement and preprocessing operations, thereby reducing power consumption while accelerating similarity searches through efficient in-memory computation
Solution Approach 2:
The patent transforms the traditional row-based memory access paradigm into column-based access, representing a dimensional change in data organization and retrieval. This columnar approach aligns with the computational requirements of similarity searches, enabling parallel processing of search vectors and significantly improving both speed and energy efficiency
2Productivity
If traditional memory architectures are used for large-scale similarity searches, then data storage capacity is sufficient, but computational complexity increases
Solution Approach 1:
The patent performs preliminary organization of data into columnar structures within the memory architecture itself. By pre-organizing data in a format optimized for similarity search operations, the system eliminates the need for complex runtime computations and data transformations, thereby reducing computational load while maintaining high search speeds
3Productivity
If column-read enabled memory is used, then similarity search speed increases, but memory architecture complexity increases
Solution Approach 1:
The column-read enabled memory architecture is designed to provide multiple functions including traditional random access, columnar access patterns, and in-memory computation capabilities. This multi-functionality allows the same memory structure to serve both as storage and as an acceleration engine for similarity searches, justifying the increased architectural complexity through substantial performance gains
Data Source
AI summary
Technologies for performing random sparse lifting and Procrustean orthogonal sparse hashing using column read-enabled memory include a device that has a memory that is column addressable and circuitry connected to the memory. The circuitry is configured to add a set of input data vectors to the memory as a set of binary dimensionally expanded vectors, including multiplying each input data vector with a projection matrix. The circuitry is also configured to produce a search hash code from a search data vector, including multiplying the search data vector with the projection matrix. Further, the circuitry is configured to determine a Hamming distance between the search hash code and each of the binary dimensionally expanded vectors.