Table 8 Comparison of running time (/ms) for point sets derived from 3D models on GT 640
From: CudaChain: an alternative algorithm for finding 2D convex hulls on the GPU
3D Model | Size | Qhull | CudaChain | Speed up | |||
|---|---|---|---|---|---|---|---|
Total | GPU | CPU | CPU (%) | ||||
Armadillo | 172k | 15 | 26.5 | 25.0 | 1.5 | 5.66 | 0.6 |
Angel | 237k | 16 | 28.0 | 26.4 | 1.6 | 5.71 | 0.6 |
Skeleton hand | 327k | 31 | 29.6 | 27.8 | 1.8 | 6.08 | 1.0 |
Dragon | 437k | 47 | 35.1 | 31.8 | 3.3 | 9.40 | 1.3 |
Happy Buddha | 543k | 62 | 42.1 | 37.4 | 4.7 | 11.16 | 1.5 |
Turbine blade | 882k | 78 | 46.3 | 42.8 | 3.5 | 7.56 | 1.7 |
Vellum manuscript | 2M | 218 | 78.5 | 75.4 | 3.1 | 3.95 | 2.8 |
Asian Dragon | 3M | 343 | 102.5 | 98.8 | 3.7 | 3.61 | 3.3 |
Thai statue | 5M | 468 | 105.5 | 101.9 | 3.6 | 3.41 | 4.4 |
Lucy | 14M | 1295 | 248.8 | 244.5 | 4.3 | 1.73 | 5.2 |