FPGNN-ATPG: An Efficient Fault Parallel Automatic Test Pattern Generator
TimeWednesday, July 13th6pm - 7pm PDT
LocationLevel 2 Lobby
Event Type
Networking Reception
Work-in-Progress Poster
DescriptionThe main challenge for parallel ATPG is to avoid test set inflation with high speedup and fault coverage. In this paper, we develop an efficient parallel ATPG, i.e., FPGNN-ATPG. Different from traditional fault collapsing approaches, we propose a graph-neural-networks-based fault classification model to figure out redundant faults, easy-to-detect faults, and hard-to-detect faults. Then, we implement fault-driven test pattern generation engines to solve them parallelly. Experimental results show GNN-based model has superior performance to classical methods. And our FPGNN-ATPG framework obtains an average of 7.56X speedup while reducing 14.13% pattern count ratio with full 100% fault coverage on an 8-core machine.