Fig.1

Illustration of GPU parallel computing

Fig.2

Data transfer method for GPU parallel computing

Fig.3

GPU computing mode

Fig.4

GPU/CPU heterogeneous computing mode

Table 1

Geometry and flow parameters of cascade

Fig.5

Grid used in cascade flow simulation

Fig.6

Static pressure contours and streamlines on suction surface of cascade blade

Fig.7

Static pressure coefficient Cp distribution at 50% blade height of cascade flow

Fig.8

Speed-up ratio of CPU program parallel computing

Table 2

Speed⁃up ratio of GPU computing mode

Fig.9

Speed-up ratio of GPU program parallel computing

Fig.10

Computational grid of T106A turbine blade

Fig.11

Vorticity contour of T106A turbine blade

Fig.12

Static pressure coefficient distribution of T106A turbine blade

Fig.13

Speed-up ratio of two heterogeneous computing modes with different numbers of CPU cores

Fig.14

Speed-up ratio of heterogeneous computing mode 1 with different numbers of GPU cards at third and fourth orders

Fig.15

Speed-up ratio of heterogeneous computing mode 2 with different numbers of GPU cards at third and fourth orders

Table 3

ANCF geometry parameters

Fig.16

Computational domain of ANCF

Fig.17

Total pressure distribution at fan duct inlet

Fig.18

Mach number distribution at fan duct inlet

Fig.19

Entropy contours at 80% blade height of ANCF

Fig.20

Pressure spectrum on duct wall of ANCF

Fig.21

First 2 BPF pressure contours at Z=-1.3D plane

Fig.22

PWLs of first 2 BPF duct modes

Fig.23

Distribution of Mach numbers at 60% blade height at different axial positions and corresponding azimuthal FFT results

Fig.24

Speed-up ratio of 4 A100 GPU cards with different numbers of CPU cores allocated per GPU