Compact Arithmetic Elements Using Low-Precision High Dynamic Range
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional computing systems inefficiently utilize transistors due to a focus on high precision arithmetic, limiting the ability to harness the full computing power of modern silicon-based processors, despite the availability of vast computational resources.
Innovation Solution
The implementation of low precision high dynamic range (LPHDR) processing elements that perform arithmetic operations with a precision of about 0.1%, allowing for a significant increase in the number of operations per unit time or unit power, using either digital logarithmic representations or analog representations to efficiently use silicon resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional high precision arithmetic is used, then computational accuracy is improved, but the number of operations per unit time or unit power deteriorates
Solution Approach 1:
The patent changes the precision parameter from conventional high precision (32-bit or 64-bit floating point) to low precision (about 0.1% precision, equivalent to roughly 10 bits). This parameter change allows significantly more arithmetic operations to be performed per unit time and per unit power, while still achieving adequate computational results for many applications through techniques like stochastic computing and probabilistic error correction
2Power
If the number of processing elements is increased, then computing power is improved, but device complexity and resource requirements worsen
Solution Approach 1:
The patent uses copying by replicating simple low-precision processing elements many times across the chip. Each processing element is identical and performs basic arithmetic operations with low precision. By copying these simple units thousands or millions of times, the system achieves high computing power without the complexity of designing and managing fewer, more complex processing elements
Solution Approach 2:
The patent segments the computing task into many independent parallel operations performed by numerous simple processing elements. Instead of having a few complex processors handle tasks sequentially or with complex coordination, the work is divided into many small independent operations that can be executed simultaneously by simple, identical processing units, reducing overall system complexity
Data Source
AI summary
Low precision computers can be efficient at finding possible answers to search problems. However, sometimes the task demands finding better answers than a single low precision search. A computer system augments low precision computing with a small amount of high precision computing, to improve search quality with little additional computing.