Recursive FPGA Document Chunking for Accurate AI Design Assistance

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

As programmable logic devices increase in complexity and size, existing design tools often provide unintelligible or erroneous results, making it difficult for designers to efficiently implement and optimize designs on integrated circuits.

Innovation Solution

A design assistant tool, such as QuartusPilot™, provides a graphical user interface and AI-assisted design analysis to identify errors, suggest optimizations, and visualize design changes, facilitating efficient design implementation on FPGAs and ASICs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If design tools are used for complex programmable logic devices, then design assistance is provided, but the results become unintelligible or erroneous

Engineering Contradiction:
Improvedesign assistanceVSAvoidresult accuracy
Core Design Contradiction:
Extent of automationVSMeasurement precision

Solution Approach 1:

The patent segments complex design documentation into hierarchical chunks (document → section → subsection → paragraph) to improve processing accuracy. This segmentation allows the system to handle large amounts of design information in manageable, context-preserving units, thereby maintaining result intelligibility while providing comprehensive design assistance for complex programmable logic devices.

Inventive Principle:
Principle #1Segmentation

2Adaptability or versatility

If design complexity increases, then more resources are referenced, but design tool results become less reliable

Engineering Contradiction:
Improveresource reference capabilityVSAvoiddesign tool reliability
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The patent implements nested contextual information structures where chunks are embedded within parent documents, sections contain subsections, and paragraphs contain sentences. This nested organization preserves contextual relationships across multiple levels of design documentation, enabling the system to reliably handle increased resource references while maintaining result accuracy even as design complexity increases.

Inventive Principle:
Principle #7Nested doll (Nesting)

3Ease of manufacture

If traditional document processing is used, then simple designs are handled, but complex designs become difficult to optimize

Engineering Contradiction:
Improvedesign implementation easeVSAvoidprogrammable logic device complexity
Core Design Contradiction:
Ease of manufactureVSDevice complexity

Solution Approach 1:

The patent transforms static document processing into a dynamic, iterative optimization process. The system continuously refines design implementations by processing chunked documentation, generating optimizations, and updating designs in multiple passes. This dynamic approach enables effective handling of complex programmable logic device designs while maintaining ease of implementation through automated iterative improvement.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS20260057161A1Systems and Methods for a Recursive Document Splitting and Chunking System for FPGA Documentation and Tool Reports
Publication Date: 2026.02.26 KOTIYAL SAURABH
  • US20260057161A1 patent drawing
  • US20260057161A1 patent drawing
  • US20260057161A1 patent drawing

AI summary

Systems or methods of the present disclosure may provide a design tool for adjusting designs implemented on programmable logic devices. The present disclosure includes receiving one or more data tables and receiving one or more design files. The one or more data tables may include technical specifications, user guides, and/or best practices. The one or more design files may include design inputs and design outputs such as timing reports, design assistant/rule violation (DRC) reports, and/or log files. The present disclosure also includes determining one or more chunks based on the one or more data tables, generating one or more embeddings based on the one or more chunks, extracting data from the one or more design files, while considering contextual information, natural break points, and cross-referencing. Furthermore, the present disclosure includes providing the one or more embeddings and the data to a large language model (LLM).