Research Article

GPU-Accelerated Parallel FDTD on Distributed Heterogeneous Platform

Table 3

Loop time information used in Figure 6.

NP Pure CPU Pure GPU CPU + GPU (cuRatio)

1 1.3541 0.1028 0.0903 (0.840)
2 0.7020 0.0535 0.0467 (0.850)
4 0.3634 0.0291 0.0279 (0.860)
8 0.2168 0.0159 0.0143 (0.860)
16 0.0923 0.0091 0.0086 (0.860)
32 0.0425 0.0055 0.0050 (0.840)
64 0.0227 0.0034 0.0033 (0.810)