Results Problem 13 (Case 1): POOSL using Simulation-Based Approach

1
Results Problem 13 (Case 1) POOSL using
Simulation-Based Approach

• Required metrics
• Throughput of B 0.370 rel. error 0.31
• Maximum occupancy of M 6
• Time-average occupancy of M 5.000 rel. error
  0.60
• Some additional metrics
• Maximum delay between successive firings of B 3
• Average delay between successive firings of
  B 2.701 rel. error 0.34
• Throughput of A 0.247 rel. error 0.32
• Maximum delay between successive firings of A 6
• Average delay between successive firings of
  A 4.051 rel. error 0.26
• Accuracy results for averages are based on
  confidence levels of 0.95
• POOSL model based on SDF representation of total
  system
• Evaluation
• Modelling effort (Specification in XML
  sdf2poosl tool) 8 min (sdf2poosl needs ltlt 1 sec)
• Simulation time (P3 _at_ 1Ghz) 16.4 sec

2
Results Problem 13 (Case 2) POOSL using
Simulation-Based Approach

• Required metrics
• Throughput of B 0.619 rel. error 0.33
• Maximum occupancy of M 6
• Time-average occupancy of M 5.673 rel. error
  0.30
• Maximum number of concurrent active threads for
  P1 2
• Time-average number of concurrent active threads
  for P1 0.742 rel. error 0.55
• Maximum number of concurrent active threads for
  P2 3
• Time-average number of concurrent active threads
  for P2 1.673 rel. error 0.53
• Some additional metrics
• Maximum delay between successive firings of B 5
• Average delay between successive firings of
  B 1.616 rel. error 0.28
• Throughput of A 0.412 rel. error 0.08
• Maximum delay between successive firings of A 5
• Average delay between successive firings of
  A 2.425 rel. error 0.09
• Accuracy results for averages are based on
  confidence levels of 0.95
• POOSL model based on SDF representation of total
  system