IP Block Scan Chain Construction via Iterative Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for constructing scan chains in integrated circuit (IC) chips face challenges in efficiently determining the optimal number and configuration of EXTEST scan chains, leading to excessively long wire lengths and inefficient testability due to the lack of a systematic approach that considers physical dimensions and component placement.
Innovation Solution
A scan chain engine is employed to iteratively partition and merge EXTEST wrapper cells within IP blocks, determining the optimal number of scan chains based on user-defined parameters, such as maximum wrapper cells and wire length, to construct physically aware EXTEST scan chains that minimize wire lengths and clock domain crossings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Length of stationary object
If scan chains are constructed without systematic partitioning and merging, then the construction process is simpler, but wire lengths become excessively long and testability efficiency deteriorates
Solution Approach 1:
The IP block is systematically partitioned into multiple regions through iterative partitioning operations. EXTEST wrapper cells are grouped into partitions based on their spatial distribution, creating manageable segments that can be independently analyzed and connected. This segmentation prevents excessively long wire paths by ensuring connections are made within localized regions rather than across the entire IP block.
Solution Approach 2:
Partitions containing EXTEST wrapper cells are selectively merged to form populated partitions. The merging process combines adjacent partitions that share common wrapper cells, creating optimized scan chain groups. This merging reduces the total number of scan chains needed while maintaining efficient wire lengths by consolidating wrapper cells into coherent groups.
2Reliability
If the number of EXTEST scan chains is increased to improve testability, then test coverage improves, but wire lengths and clock domain crossings increase
Solution Approach 1:
The patent applies different treatment to different regions of the IP block based on the local distribution of EXTEST wrapper cells. Regions with high density of wrapper cells are merged into populated partitions, while regions with sparse distribution are handled differently. This local quality approach ensures that scan chains are optimized for their specific regional characteristics, achieving good testability without uniformly increasing wire lengths across the entire block.
Solution Approach 2:
The patent transforms the one-dimensional problem of scan chain count into a two-dimensional optimization problem by considering both the number of scan chains and their spatial arrangement. Through iterative partitioning and merging operations, the system optimizes the spatial configuration of scan chains, achieving efficient testability while controlling wire lengths by exploiting the spatial dimension of the IP block layout.
3Length of stationary object
If partitioning is applied to reduce wire lengths, then the number of partitions increases, but the construction process becomes more complex
Solution Approach 1:
The patent performs preliminary partitioning of the IP block into regions before constructing scan chains. By pre-organizing wrapper cells into partitions based on their spatial locations, the system prepares the data structure in advance to facilitate efficient scan chain construction. This preliminary action reduces the complexity of the subsequent merging and connection operations, as the partitioning has already grouped wrapper cells in a way that minimizes wire lengths.
Solution Approach 2:
The iterative partitioning and merging process incorporates feedback mechanisms where the results of each partitioning operation inform subsequent merging decisions. The system evaluates the characteristics of generated partitions (such as the number of wrapper cells and spatial distribution) and uses this feedback to determine which partitions to merge next. This feedback-driven approach automates the complex construction process, reducing manual intervention while achieving optimal wire length reduction.
Data Source
AI summary
A scan chain engine can determine a set number of EXTEST scan chains for the IP block and based on a predetermined maximum number of EXTEST wrapper cells per EXTEST scan chain. The scan chain engine iteratively executes partitioning on the IP block to generate a set of partitions. Each partition in the set of partitions has a number of EXTEST wrapper cells that does not exceed the maximum number of EXTEST wrapper cells per EXTEST scan chain. The scan chain engine selectively merges partitions of the set of partitions to form a set of populated partitions that each include an EXTEST wrapper cell. The number of partitions is equal to the set number of EXTEST scan chains for the IP block. The scan chain engine generates wire paths connecting EXTEST wrapper cells of each populated partition to construct the set number of EXTEST scan chains for the IP block.