Collaborative Conversational Agents for IC Design Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional IC design methods face challenges in efficiently achieving optimal power, performance, area, and cost (PPAC) metrics due to the lack of transparency in EDA tools' internal workings and the need for redundant experiments, leading to long turnaround times and no guarantee of PPAC improvements.
Innovation Solution
Implementing collaborative conversational agents that communicate using a shared vocabulary to adaptively modify design parameters, allowing for one-shot PPAC optimization by providing feedback to designers on essential changes, and leveraging AI for metric prediction to prioritize design options.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional EDA tools are used with full reruns of tool commands, then PPAC optimization can be achieved, but turnaround time increases significantly
Solution Approach 1:
The patent applies preliminary action by running a reduced set of tool commands before full optimization to establish an initial design state. This preliminary execution provides enough information to guide subsequent selective reruns, avoiding the need to always execute complete tool flows from scratch and thereby reducing overall turnaround time while maintaining optimization reliability.
Solution Approach 2:
The patent implements feedback mechanisms where the system monitors PPAC metrics from previous tool executions and uses this information to intelligently determine which reruns are necessary. This feedback loop allows the system to avoid redundant full reruns by leveraging results from partial executions, thus reducing turnaround time without compromising optimization effectiveness.
2Measurement precision
If multiple versions of models and silicon data are maintained, then design accuracy is improved, but difficulty in identifying the best version increases
Solution Approach 1:
The patent introduces an intermediary system that manages and coordinates between multiple versions of models and silicon data. This intermediary layer abstracts the complexity of version selection and data management from the designer, providing a unified interface that automatically handles version identification and selection based on design requirements, thereby maintaining accuracy while reducing the difficulty of version management.
3Reliability
If redundant experiments are executed, then PPAC optimization is ensured, but designer effort and time consumption increase
Solution Approach 1:
The patent applies partial action by executing only the necessary subset of tool commands and experiments required for PPAC optimization rather than always running complete redundant sets. The system determines the minimal sufficient action set based on current design state and PPAC goals, reducing designer effort and time consumption while maintaining adequate optimization reliability through intelligent selection of critical experiments.
Data Source
AI summary
Systems, apparatuses and methods may provide for technology that determines a vocabulary based on EDA tool terminologies and/or a natural language, queries and recommends, by a plurality of virtual agents, actions based on a design state and the vocabulary, wherein the plurality of agents is to include a tool agent and a designer agent, and executes a set of modifications to the design state in accordance with a collaboration between the plurality of agents. The technology may also convert a first user query from a first format to a second format, wherein the first format is incompatible with a trained AI model of a hardware architecture and the second format is compatible with the trained AI model, generate one or more predictions from the trained AI model based on the converted first user query, and select a subset of recommendations from a set of candidate architectures based on the prediction(s).