In-Memory Bit String Computing With Posit Format Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current memory systems face limitations in performing bit string operations due to finite memory resources and the need for higher precision and accuracy, particularly in computing systems that rely on floating-point formats, which can lead to overflow, underflow, and reduced performance in applications like machine learning.
Innovation Solution
The implementation of bit string operations using a universal number format, specifically the posit format, which allows for higher precision, dynamic range, and accuracy by converting bit strings from floating-point format to posit format within memory arrays, enabling arithmetic and logical operations without transferring data externally, thus improving processing speed and reducing memory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If floating-point format is used for bit string operations, then compatibility with existing systems is maintained, but precision and accuracy are reduced leading to overflow and underflow issues
Solution Approach 1:
The patent changes the numerical format parameter from floating-point to posit format. This parameter change enables the system to achieve higher precision and eliminate overflow/underflow issues while maintaining the same bit width, directly resolving the technical contradiction between precision and reliability.
2Productivity
If bit string operations are performed within memory arrays, then processing speed and memory efficiency are improved, but device complexity increases
Solution Approach 1:
The patent merges the computing function with the memory array by enabling bit string operations to be performed directly within the memory device. This integration allows data to be processed in-place without external transfer, improving processing speed while the complexity is managed through unified architecture design.
Solution Approach 2:
The memory device is designed to perform multiple functions: both data storage and bit string operations. By making the memory device universal, it can handle both traditional memory functions and computational tasks, improving productivity while distributing complexity across a multi-functional system.
3Measurement precision
If higher precision operations are implemented, then accuracy is improved, but memory resources are consumed faster
Solution Approach 1:
The patent changes the numerical representation parameter to posit format, which achieves higher precision with the same bit width. This allows accurate operations to be performed using existing memory resources without requiring additional memory capacity, resolving the contradiction between accuracy and memory resource consumption.
Data Source
AI summary
Systems, apparatuses, and methods related to bit string operations in memory are described. The bit string operations may be performed within a memory array without transferring the bit strings or intermediate results of the operations to circuitry external to the memory array. For instance, sensing circuitry that can include a sense amplifier and a compute component can be coupled to a memory array. A controller can be coupled to the sensing circuitry and can be configured to cause one or more bit strings that are formatted according to a universal number format or a posit format to be transferred from the memory array to the sensing circuitry. The sensing circuitry can perform an arithmetic operation, a logical operation, or both using the one or more bit strings.