4/26/05Han: ELEC72501 Department of Electrical and Computer Engineering Auburn University, AL K.Han Development of Parallel Distributed Computing System for ATPG Program and Evaluation of the speedup
4/26/05Han: ELEC72502 Develop data parallel distributed computing system to run ATPG program on a distributed system of computers Evaluate the speedup as a function of the number of processors Develop data parallel distributed computing system to run ATPG program on a distributed system of computers Evaluate the speedup as a function of the number of processors The Objective of Project By K. Han
4/26/05Han: ELEC Partition the original fault list (circuit.fault) into piece 2.Distribute each sub-list to an instance of fault simulator and test generator 3.Execute ATPG program (HITEC/PROOFS) on each partitioned fault list separately and simultaneously. 4.Collect to produce the final result (circuit.atp) for test generation 5.Combine all circuit.atp files into one file 6.Eliminate all the duplicate entries from this file 1.Partition the original fault list (circuit.fault) into piece 2.Distribute each sub-list to an instance of fault simulator and test generator 3.Execute ATPG program (HITEC/PROOFS) on each partitioned fault list separately and simultaneously. 4.Collect to produce the final result (circuit.atp) for test generation 5.Combine all circuit.atp files into one file 6.Eliminate all the duplicate entries from this file The Objective of Project (1) By K. Han
4/26/05Han: ELEC Cluster Network System (In Aerospace Computer Lab) - Compiler including MPI library function -Cluster Network System (In Aerospace Computer Lab) - Compiler including MPI library function The Requirements By K. Han
4/26/05Han: ELEC72505 HITEC/PROOFS Master slave ……. Structure By K. Han slave HITEC/PROOFS
4/26/05Han: ELEC72506 Eliminate all the duplicate test Master Slaves Levelize circuit Generate faultlist equiv dominators testgen Combine circuit.atp into one file Circuit.atp Circuit.fault]
4/26/05Han: ELEC72507 From Sequential System e+02 139 sec np = 4 100 sec np = 8 59 sec np = 16 53 sec Comparison of Execution Time By K. Han
4/26/05Han: ELEC72508 Speedup By K. Han Speedup Number of Nodes
4/26/05Han: ELEC72509 Prospective improvement By K. Han -Load balancing - More optimized algorithm -Load balancing - More optimized algorithm
4/26/05Han: ELEC Speedup (Save execution time) -More convenient and cost-effective -Speedup (Save execution time) -More convenient and cost-effective The Development By K. Han