Table 1 The experimental performance results for cloth simulation
From: Plausible mass-spring system using parallel computing on mobile devices
# of nodes | 2500 | 4900 | 10,000 | 40,000 | 200,000 | 350,000 |
# of springs | 29,004 | 57,404 | 118,004 | 476,004 | 2,391,004 | 4,188,004 |
Semi-Euler integration (ms) | ||||||
 CPU only | 19.475 | 41.890 | 80.480 | 331.933 | 1642.693 | 2733.484 |
 Multi-thread CPU | 6.60 | 13.12 | 21.98 | 76.10 | 349.11 | 586.35 |
 GPU | 1.827 | 1.963 | 2.791 | 20.264 | 67.130 | 115.015 |
Midpoint integration (ms) | ||||||
 CPU only | 20.377 | 44.047 | 84.103 | 345.238 | 1719.711 | 2852.775 |
 Multi-thread CPU | 7.03 | 13.50 | 23.59 | 79.02 | 368.26 | 629.23 |
 GPU | 1.735 | 1.899 | 2.765 | 21.986 | 68.454 | 117.237 |
4th-order Runge-Kutta integration (ms) | ||||||
 CPU only | 23.954 | 50.95 | 97.763 | 399.049 | 1922.532 | 3335.518 |
 Multi-thread CPU | 7.94 | 15.39 | 26.70 | 92.71 | 453.14 | 778.48 |
 GPU | 1.688 | 1.793 | 2.800 | 22.393 | 68.117 | 115.649 |