Instruction Encoding With Predictive Word Omission for Code Density

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing encoding and decoding methods for instruction sets face challenges in achieving high code density while maintaining flexibility and reducing hardware complexity, particularly when adding new instructions or handling variable length instructions, which can lead to inefficiencies and errors.

Innovation Solution

A method that rearranges instruction bits into a variable number of fixed-length or multiple of fixed-length words, truncating or masking words with predictable values to achieve compression, and uses empirical data and simulation to determine mappings and predicted values for efficient encoding and decoding.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If variable length instruction words are used to increase code density, then memory requirements are reduced, but hardware complexity increases due to the need to handle different field omissions and identifier interpretations

Engineering Contradiction:
Improvememory requirementsVSAvoidhardware complexity
Core Design Contradiction:
Quantity of substanceVSDevice complexity

Solution Approach 1:

The instruction word is segmented into fixed fields with predetermined bit lengths, where each field has a specific function. By dividing the instruction into structured segments (opcode, operand fields, etc.), the decoder can systematically process each segment without requiring complex variable-length field parsing logic, thus reducing hardware complexity while maintaining compression efficiency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the parameter of instruction word length from variable to fixed, while achieving compression through selective field omission rather than variable word length. This parameter change simplifies the hardware decoder design as it no longer needs to dynamically determine field boundaries based on variable lengths, while still achieving code density improvement through the identifier mechanism.

Inventive Principle:
Principle #35Parameter changes

2Quantity of substance

If fields are omitted from instructions to compress bit length, then code density increases, but reliability decreases due to potential errors in identifying omitted fields

Engineering Contradiction:
Improvecode densityVSAvoiddecoding accuracy
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The identifier field provides feedback information to the decoder about which fields are omitted. This feedback mechanism allows the decoder to accurately reconstruct the original instruction structure by knowing exactly which fields should be present and which are omitted, thereby maintaining high decoding reliability while achieving compression through field omission.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The identifier acts as an intermediary between the encoded instruction and the decoder. It carries metadata information about the instruction structure (which fields are omitted) without being part of the operational instruction itself. This intermediary enables the decoder to correctly interpret the compressed instruction format and reliably reconstruct the intended operation.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Adaptability or versatility

If manual generation of encoding/decoding software and hardware is performed, then adaptability to processor variants is improved, but time consumption and error probability increase

Engineering Contradiction:
Improveadaptability to processor variantsVSAvoiddevelopment time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent employs automated generation tools that self-generate the encoding software and decoding hardware descriptions from processor architecture specifications. This self-service approach eliminates the need for manual, error-prone generation processes while maintaining full adaptability to different processor variants. The automated tools can systematically generate correct and consistent encoding/decoding implementations for any processor variant by interpreting the architectural specifications.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical/manual process of generating encoding and decoding implementations with an automated computational system. Instead of manually writing and debugging encoder/decoder code for each processor variant, the system uses automated generation tools that computationally produce the correct implementations, significantly reducing development time and error probability while maintaining adaptability.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP3754484B1Generating encoding software and decoding means
Publication Date: 2025.10.22 IMAGINATION TECH LTD
  • EP3754484B1 patent drawingFigure 1
  • EP3754484B1 patent drawingFigure 2
  • EP3754484B1 patent drawingFigure 3

AI summary

Methods of encoding and decoding are described which use a variable number of instruction words to encode instructions from an instruction set, such that different instructions within the instruction set may be encoded using different numbers of instruction words. To encode an instruction, the bits within the instruction are re-ordered and formed into instruction words based upon their variance as determined using empirical or simulation data. The bits in the instruction words are compared to corresponding predicted values and some or all of the instruction words that match the predicted values are omitted from the encoded instruction.