Variable-Precision Floating-Point Storage for Memory Footprint Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing variable precision computing methods face challenges with memory footprint inefficiencies, leading to significant computational inaccuracies due to rounding, cancellation, and absorption errors, particularly in numerical instability and high-precision requirements.
Innovation Solution
A method and device for storing variable precision floating-point values by reducing their bit length to a size limit, using a load and store circuit to convert and truncate values, and storing them in memory zones with programmable size limits, allowing dynamic bit length adjustment and efficient memory usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If variable precision floating point values are stored with their original bit length, then computational precision is maintained, but memory footprint increases and access efficiency decreases
Solution Approach 1:
The patent segments the storage space into multiple zones with different size limits, allowing variable precision floating point values to be stored in the most efficient zone. This segmentation enables the system to balance between maintaining computational precision and optimizing memory access efficiency by placing values in appropriately sized storage regions.
Solution Approach 2:
The patent changes the parameter of bit length by reducing variable precision floating point values to fit within size limits of different storage zones. This parameter transformation allows the system to optimize memory footprint and access efficiency while maintaining sufficient computational precision for the intended applications.
2Productivity
If bit length is reduced to fit size limits, then memory usage is optimized, but computational precision may be lost
Solution Approach 1:
The patent implements dynamic precision management where the bit length of floating point values is adjusted based on the size limit of the target storage zone. This dynamic approach allows the system to optimize memory usage by reducing precision only when and where it is acceptable, while maintaining higher precision in zones where it is still needed for computational accuracy.
3Ease of manufacture
If fixed size storage zones are used, then memory allocation is simplified, but flexibility in storing different precision requirements is reduced
Solution Approach 1:
The patent creates a universal storage system with multiple zones that can handle different precision requirements through a common architecture. Each zone is designed with a size limit that can accommodate specific precision needs, allowing the system to serve multiple precision requirements while maintaining simple fixed-size zone allocation. This multi-functional design enables both simplified memory allocation and adaptable precision storage.
Data Source
AI summary
The present disclosure relates to a method of storing, by a load and store circuit or other processing means, a variable precision floating point value to a memory address of a memory, the method comprising: reducing the bit length of the variable precision floating point value to no more than a size limit, and storing the variable precision floating point value to one of a plurality of storage zones in the memory, each of the plurality of storage zones having a storage space equal to or greater than the size limit (MBB).