Page 56 - Data Science Algorithms in a Week
P. 56

40                   Edwin Cortes, Luis Rabelo and Gene Lee

                              14. STD processor Speed: It is the standard deviation of the speed of the CPUs
                                 used in the simulation.
                              15. Mean RAM: It is the mean of the RAM memory used by the CPUs in the
                                 system.
                              16. STD RAM: It is the standard deviation of the RAM memory used by the
                                 CPUs in the system.
                              17. Critical  Path%:  It  is  the  Critical  Path  taking  into  consideration  the
                                 sequential estimated processing time.
                              18. Theoretical  Speedup:  It  is  the  theoretical  (maximum)  speedup  to  be
                                 achieved with perfect parallelism in the simulation.
                              19. Local Events/(Local Events + External Events): It is the ratio of the total
                                 local events divided by the summation of the total local events and the total
                                 external events during a specific unit of Simulation Time (estimated).
                              20. Subscribers/(Publishers + Subscribers): It is the ratio of the total number
                                 of objects subscribing to a particular object divided by the summation of the
                                 total number of publishers and subscribers.
                              21. Block  or  Scatter?:  Block  and  scatter  are  decomposition  algorithms  being
                                 used to distribute the simulation objects in the parallel/distributed system - If
                                 Block is being selected then this value is 1. and if Scatter is selected then this
                                 value is 0.

                          For example, for the discussed aircraft detection implementation, this is the following
                       input vector using the hardware and complexity specifications from Figures 6 and 7, and
                       Tables  1  and  2  for  a  configuration  of  4  Global  Nodes  and  1  Local  Node  (a  loosely
                       coupled system) using “Block” as the distribution scheme for the simulation objects is
                       shown in Table 3.
                          And the output for a DBN will be based on Table 4 where the Wall Clock Time for
                       BTW is 11.4 seconds, for BTB is 162.6 seconds, and for TW is 10.9 seconds. Table 4
                       displays the output vector of the respective case study of aircraft detection.

                        Table 3: Vector that defines the PDDES implementation for the aircraft detection
                                      with 4 Global Nodes and 1 Local Node using Block


                           Complexity Parameters that Capture the hardware/software Structure of a Parallel
                                              Distributed Discrete-Event Simulator
                        Total Simulation Program Cognitive Weights                            2919
                        Number of Sim objects                                                   6
                        Types of Sim objects                                                    3
                        Mean Events per Object                                                  1
                        STD Events per Simulation Object                                        0
                        Mean Cog Weights of All objects                                       1345
                        STD Cog Weights of All objects                                        1317
   51   52   53   54   55   56   57   58   59   60   61