Logic Design Numeric Format Flexibility for Area and Power Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing technologies face challenges in optimizing numeric formats for integrated circuits and software, leading to inefficiencies in performance, area consumption, and power usage due to rigid format specifications that interfere with other optimizations and do not adapt to changing requirements.
Innovation Solution
The method involves interpreting numeric format directives as minimum specifications for range and accuracy, allowing for increased flexibility in word lengths and formats within data paths, deferring format reductions when necessary to enable logic optimizations, and automatically switching to more power-efficient formats like sign magnitude when power limits are exceeded.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If format directives are treated as exact specifications, then implementation cost and area consumption are reduced, but logic optimization opportunities are lost and accuracy may deteriorate
Solution Approach 1:
The system dynamically adjusts numeric format specifications during synthesis based on actual optimization opportunities. Format directives are treated as flexible guidelines rather than fixed constraints, allowing the synthesis tool to select optimal formats that balance area consumption with accuracy requirements and enable logic optimizations like carry-save adder merging.
Solution Approach 2:
The invention changes the interpretation parameter of format directives from 'exact specification' to 'minimum requirement'. This parameter change allows the synthesis tool to exceed specified formats when beneficial for optimization, while still meeting the minimum accuracy and area constraints defined by the directives.
2Use of energy by stationary object
If format reductions are applied early in data paths, then area and power consumption are reduced, but logic optimizations such as operator merging are prevented
Solution Approach 1:
The system performs logic optimizations like operator merging at an early stage before final format reduction is applied. By postponing format reduction until after optimizations are complete, the synthesis tool can identify and merge operators that would otherwise be fragmented by early format constraints, thereby reducing overall power consumption more effectively.
Solution Approach 2:
The format specification is made dynamic across different stages of synthesis. Early stages allow flexible format interpretations that enable optimizations, while later stages apply format reductions to meet area and power constraints. This dynamic adjustment resolves the contradiction between early reduction benefits and optimization flexibility.
3Device complexity
If fixed format specifications are used, then implementation complexity is reduced, but performance and area optimization are limited
Solution Approach 1:
The synthesis tool automatically manages format optimizations without requiring complex manual format specifications from the designer. The tool services itself by intelligently adjusting formats based on optimization opportunities, thereby reducing the burden on the designer while achieving superior performance and area efficiency.
4Ease of manufacture
If standard numeric formats are used throughout, then ease of implementation is improved, but power consumption and area usage increase
Solution Approach 1:
The system applies different numeric format optimizations to different parts of the data path based on local requirements. Rather than uniformly applying standard formats throughout, the synthesis tool identifies specific locations where format adjustments can reduce power consumption and area usage while maintaining overall system functionality and accuracy.
Data Source
AI summary
Methods and apparatuses for designing logic are described. In one embodiment, a directive which specifies a numeric format for data in a data processing operation in a logic design is determined. The directive is used as a minimum format, rather than an exact or required format to create or change at least a portion of a representation of logic in the logic design to perform the data processing operation. Other methods are disclosed, and systems and machine readable media are also disclosed.


