EDA Data Collection Platform Using Key-Value Logging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current electronic design automation (EDA) tools generate rich, domain-specific data that is often discarded or stored in log files, with no standard format, making data analysis and mining for business insights time-consuming and non-scalable due to the lack of open-source tools capable of handling such formatted data.
Innovation Solution
A learning-ready platform is developed with a uniform data collection mechanism using key-value pairs and a high-performance data processing system for real-time data integration, capable of automatically collecting and processing data from EDA tools, including log-file parsing for non-standard formats.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If conventional log file parsing methods are used to collect data from EDA tools, then data can be extracted from existing logs, but the process is error-prone, highly sensitive to log format changes, and not scalable
Solution Approach 1:
The patent introduces an intermediary component (log parser or data collector) that sits between the EDA tools and the data storage system. This intermediary handles the complexity of log format variations and parsing errors, providing a stable interface for data collection. The intermediary translates various EDA tool log formats into a standardized internal representation, isolating the rest of the system from format changes and improving both automation reliability and scalability.
2Adaptability or versatility
If data is stored in log files with no standard format, then all EDA tool data can be captured, but data analysis and mining become time-consuming and non-scalable
Solution Approach 1:
The patent segments the data collection and processing system into distinct modular components: data collectors for different EDA tools, format parsers, data validation layers, and analysis modules. Each component handles specific data formats independently, allowing the system to maintain adaptability to various log formats while improving productivity through specialized, optimized processing pipelines for each data type.
Solution Approach 2:
The patent transforms unstructured or semi-structured log file data into structured data with standardized parameters and schemas. By changing the parameter representation from free-form text to defined data structures with consistent formatting, the system maintains compatibility with diverse EDA tool outputs while enabling efficient, scalable analysis through standardized query and processing operations.
3Measurement precision
If home-grown solutions are built for each EDA tool phase, then specific phase data can be addressed, but the solution targets at most one phase and lacks cross-phase capability
Solution Approach 1:
The patent creates a universal data collection and processing platform that can handle multiple EDA tool phases (synthesis, placement, routing, timing analysis, verification, etc.) through a single integrated system. The platform uses phase-agnostic data collection mechanisms that work across all EDA tool types, while maintaining the ability to extract phase-specific details when needed. This eliminates the need for separate home-grown solutions for each phase while preserving precise phase-specific data capture capabilities.
Data Source
AI summary
A system and method for operating a high-performance learning-ready platform for real-time analytics involves operating a collector to collect logging data from electronic design automation (EDA) tools. The collector operates either (A) a set of programming interface (i.e., function calls) integrated with an EDA tool, or (B) a selector to extract the logging data from EDA tool logs. The collector generates data logs in a key-value pair data logging format from the logging data. A memory controller loads and indexes the data logs into a controlled memory data structure. A comparator in an archiving engine prioritizes storage of particular data logs based, in part, on previously stored data logs in the controlled memory data structure.