Speculative Scheduling for ATPG Multi-Core Runtime
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As chip design sizes grow, automatic test pattern generation (ATPG) and design-for-test (DFT) processes become computationally intensive, requiring longer simulation times and more runs due to increased parameter space exploration, which is limited by test hardware runtime, necessitating a speedup in multi-core computing environments.
Innovation Solution
The method employs sequential multi-level parameter value optimization with speculative scheduling to parallelize test compression runs, allowing early termination of non-optimal runs and maximizing the use of available cores, thereby optimizing fault coverage metrics and reducing runtime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If sequential parameter optimization runs are executed on multi-core nodes, then productivity is improved through parallel execution, but device complexity increases due to speculative scheduling and run management
Solution Approach 1:
The system performs preliminary actions by speculatively scheduling and initiating parameter optimization runs at multiple levels before knowing the optimal parameter values. Multiple sequential runs are launched in advance on different cores, and the system waits for results to determine which runs to continue or terminate. This allows the system to maximize core utilization while managing complexity through a structured approach to parallel execution.
2Productivity
If more parameter search runs are launched in parallel on multi-core nodes, then productivity increases, but loss of time increases due to overhead from managing multiple concurrent runs
Solution Approach 1:
The system applies partial action by launching a limited number of parallel runs at each level rather than exhaustively exploring all parameter combinations simultaneously. The speculative scheduling approach initiates runs with assumed parameter values, and if those assumptions prove incorrect, the runs are terminated early. This prevents excessive resource consumption while maintaining high productivity through selective parallel execution.
3Measurement precision
If comprehensive parameter space exploration is performed, then measurement precision is improved for fault coverage metrics, but duration of action increases due to extended simulation time
Solution Approach 1:
The system performs preliminary evaluations at each parameter level to determine which parameter values warrant further exploration. By speculatively scheduling runs with assumed optimal values and terminating runs that prove suboptimal, the system efficiently allocates simulation time to the most promising parameter combinations. This maintains measurement precision for fault coverage while significantly reducing overall simulation duration compared to exhaustive exploration.
Solution Approach 2:
The system dynamically adjusts the exploration strategy based on runtime results. As parameter optimization runs complete, the system learns which parameter values and levels are most promising and reallocates computational resources accordingly. This dynamic approach allows comprehensive exploration of critical parameter spaces while avoiding wasted simulation time on unpromising configurations, thereby improving fault coverage measurement precision without proportionally increasing runtime.
Data Source
AI summary
Systems and methods provide acceleration of automatic test pattern generation in a multi-core computing environment via multi-level parameter value optimization for a parameter set with speculative scheduling. The methods described herein use multi-core based parallel runs to parallelize sequential execution, speculative software execution to explore possible parameter sets, and terminate/prune runs when the optimum parameter value is found at a previous level. The present invention evaluates the design prior to the implementation of the compression IP so that it can define the configuration of DFT and ATPG to maximize the results of compression as measured by test data volume and test application time.