
High-High-PPerformance Optimization Methodserformance Optimization Methods 87
400, 800, 1600, 3200 and 6400. The results for F1, F2 and F5 are shown in
T
ables 3.5,
3.6 and 3.7, respectively. Figure 3.5 illustrates for the test function
F5, shows population size vs. the execution time for the three different
implementations: sequential , parallel using multicore technology, and for
the CUDA.
Table 3.3 Results of the GA programmed in sequential mode. Four variables were used for
all the test functions.
Number of
generations
F1 F2 F5
Time
(ms)
Objective
value
Time
(ms)
Objective
value
Time (ms) Objective
value
10 47.00 972.55 47.00 2433.11 78.00 2921.77