Vector Bitwise Rotations Using Shared Byte-Wise Mux Logic
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vector bitwise rotation instructions require additional dedicated logic for each supported element size, which is costly in terms of area and power consumption.
Innovation Solution
A combination of a byte-wise anything-to-anything multiplexer (mux) and bitwise shifts is used to perform vector bitwise rotations, leveraging existing muxes in CPUs and reducing the need for dedicated rotate logic.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If dedicated bitwise reverse logic is implemented for each supported element size, then vector bitwise rotation can be performed, but the area cost increases significantly
Solution Approach 1:
The byte-wise mux is designed to perform multiple functions: it can perform byte-wise permutations for vector bitwise rotation, and it can also serve as a general-purpose multiplexer for other CPU operations. This multi-functionality eliminates the need for dedicated bitwise reverse logic for each element size, reducing the overall area cost while maintaining rotation capability.
Solution Approach 2:
The invention merges the vector bitwise rotation function with the existing byte-wise mux structure. Instead of implementing separate dedicated logic for rotation, the rotation functionality is combined with the general-purpose mux, allowing both operations to share the same hardware resources and reducing total area consumption.
2Productivity
If dedicated bitwise reverse logic is implemented for each supported element size, then vector bitwise rotation can be performed, but power consumption increases
Solution Approach 1:
The byte-wise mux serves as a universal component that handles both general-purpose multiplexing operations and specific vector bitwise rotation operations. This eliminates the need for multiple dedicated logic blocks that would each consume power, replacing them with a single multi-functional unit that consumes less total power.
Solution Approach 2:
The rotation functionality is merged with the general-purpose mux, combining what would have been separate power-consuming logic blocks into a single shared hardware structure. This merging reduces the total power consumption by eliminating redundant circuitry and shared control logic.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A combination of a byte-wise anything-to-anything mux and one or more bitwise shifts is used to perform vector bitwise rotations, with even and odd elements of the vector operated on separately.