Soft Float Function Circuit for Low-Power SoC Floating Point
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing system-on-chips (SoCs) face increased power consumption and performance degradation when performing floating point operations due to the inclusion of hardware for these operations, which is not suitable for low-power IoT devices.
Innovation Solution
Incorporating a soft float function circuit that performs floating point operations without hardware, using a CPU for integer operations and a separate soft float function circuit for floating point operations, with a controller managing multiple floating point units and storage devices to optimize performance and power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If hardware floating point operation units are included in the CPU, then floating point operation performance is improved, but power consumption increases
Solution Approach 1:
The system is divided into two independent parts: a CPU that performs only integer operations and a separate soft float function circuit that performs floating point operations. This segmentation allows the CPU to remain power-efficient while the floating point operations are handled by a dedicated circuit that only activates when needed, resolving the contradiction between performance and power consumption.
Solution Approach 2:
The floating point operation functionality is extracted from the CPU and placed in a separate soft float function circuit. This extraction allows the CPU to operate at lower power consumption levels while the floating point operations are performed by the extracted circuit, which can be powered on only when floating point operations are required.
2Productivity
If hardware floating point operation units are included in the CPU, then floating point operation capability is improved, but device complexity increases
Solution Approach 1:
The system architecture is segmented into a simple CPU for integer operations and a separate soft float function circuit for floating point operations. This segmentation simplifies the CPU structure by removing unnecessary floating point hardware while maintaining floating point capability through the separate circuit, thus reducing overall device complexity.
Solution Approach 2:
The floating point operation capability is extracted from the CPU structure and placed in a separate circuit. This extraction simplifies the CPU design by eliminating complex floating point hardware, reducing the CPU's structural complexity while preserving floating point functionality through the extracted soft float function circuit.
3Use of energy by moving object
If CPU performs floating point operations in software manner, then power consumption is reduced, but operation speed decreases
Solution Approach 1:
A soft float function circuit acts as an intermediary between the CPU and the floating point operation requirements. This intermediary circuit receives commands from the CPU and executes floating point operations at high speed, bridging the gap between the CPU's low-power software execution and the need for fast floating point operations.
Solution Approach 2:
The system segments the processing tasks: the CPU handles control logic and integer operations at low power consumption, while the soft float function circuit handles floating point calculations at high speed. This segmentation allows each component to optimize for its specific function, achieving both low power consumption and high operation speed.
Data Source
AI summary
Disclosed is an SoC including a CPU that generates a first function signal including a first command for a first soft float function while not having a floating point operation function, a system bus, and a soft float function circuit that receives the first function signal from the CPU through the system bus, and performs a first floating point operation corresponding to the first soft float function based on the first command.