Difference between revisions of "Hardware"
From Agisoft
(→Benchmark) |
(→Benchmark) |
||
Line 53: | Line 53: | ||
| ~1360 | | ~1360 | ||
| ~1475 | | ~1475 | ||
− | | 1.2.0 build 2172 ( | + | | 1.2.0 build 2172 (Windows 7) |
| NVIDIA 358.50 | | NVIDIA 358.50 | ||
|- | |- | ||
Line 62: | Line 62: | ||
| ~880 | | ~880 | ||
| ~917 | | ~917 | ||
− | | 1.2.0 build 2172 ( | + | | 1.2.0 build 2172 (Windows 7) |
| NVIDIA 353.62 | | NVIDIA 353.62 | ||
|- | |- | ||
Line 89: | Line 89: | ||
| ~ 635 | | ~ 635 | ||
| ~ 637 | | ~ 637 | ||
− | | 1.2.0 build 2172 ( | + | | 1.2.0 build 2172 (Windows 7) |
| NVIDIA 358.50 | | NVIDIA 358.50 | ||
|- | |- | ||
Line 124: | Line 124: | ||
| ~ | | ~ | ||
| ~ 977 | | ~ 977 | ||
− | | ~ | + | | ~ 1012 |
| 1.2.0 build 2198 (Windows 10) | | 1.2.0 build 2198 (Windows 10) | ||
| | | |
Revision as of 10:56, 23 November 2015
PhotoScan Memory Requirements
GPU Comparison
Agisoft Sample Data
- Building (50 images, 12MPix)
- Doll (61 images, 14MPix)
- Monument (32 images, 12Mpix)
- Spherical panorama (73 images, 24MPix)
Benchmark
GPU | Dataset | Medium* | High* | Ultra* | Agisoft Version | Driver Version |
---|---|---|---|---|---|---|
Radeon R9 290X | ~ | ~1059 | ~1108 | |||
Radeon R9 290 | ~ 698.5 | ~ 905 | ~ | |||
GeForce GTX TITAN X | building sample | ~1000 | ~1352 | ~1303 | ||
GeForce GTX TITAN X | monument sample | ~1060 | ~1360 | ~1475 | 1.2.0 build 2172 (Windows 7) | NVIDIA 358.50 |
GeForce GTX TITAN | monument sample | ~740 | ~880 | ~917 | 1.2.0 build 2172 (Windows 7) | NVIDIA 353.62 |
GeForce GTX 980 | ~ | ~1005 | ~1012 | |||
GeForce GTX 970 | ~ 854 | ~ 843 | ~ | |||
GeForce GTX 680 | monument sample | ~ 570 | ~ 635 | ~ 637 | 1.2.0 build 2172 (Windows 7) | NVIDIA 358.50 |
Quadro K5000 | ~ | ~ 426 | ~ 438 | |||
Quadro M6000 | monument sample | ~ | ~1630 | ~1658 | 1.2.0 build 2181 (Ubuntu 14.04) | NVIDIA 346.96 |
FirePro W9100 | monument sample | ~ | ~ | ~ 1031 | 1.2.0 build 2172 (Ubuntu 14.04) | |
FirePro W9100 | monument sample | ~ | ~ 977 | ~ 1012 | 1.2.0 build 2198 (Windows 10) |
/* - values are given in millions samples per second (check Console pane)
/** - for benchmarking it is recommended to disable all CPU cores and leave only one GPU active in OpenCL tab.