Table 1 Execution time (/ms) of CPU and GPU versions of the AIDW algorithm on single precision
From: Improving GPU-accelerated adaptive IDW interpolation algorithm using fast kNN search
Version | Data size (1KÂ =Â 1024) | ||||
|---|---|---|---|---|---|
10K | 50K | 100K | 500K | 1000K | |
CPU/serial | 6791 | 168,234 | 673,806 | 16,852,984 | 67,471,402 |
Original naive version | 65.3 | 863 | 2884 | 63,599 | 250,574 |
Original tiled version | 61.3 | 714 | 2242 | 43,843 | 168,189 |
Improved naive version | 27.9 | 400 | 1366 | 31,306 | 124,353 |
Improved tiled version | 21.0 | 233 | 771 | 16,797 | 66,338 |