EN
Guide
GPU benchmark
Darius Rückert
- last updated on Dec 12, 2024
The QRT multi-GPU benchmark evaluates the Voxray QRT reconstruction performance of datasets taken with 1996x2000, 3008x3008 and 4096x4096 detectors on various GPU configurations to help you choose the right GPU for your needs.
Benchmark for reconstructing a dataset taken with a 1996x2000 detector:
| Reconstruction Time | |
|---|---|
1 x RTX 4090 | 0:32h |
2 x RTX 4090 | 0:18h |
1 x RTX 4500 Ada | 1:05h |
2 x RTX 4500 Ada | 0:34h |
4 x RTX 4500 Ada | 0:20h |
| Dataset Overview | |
|---|---|
CT Scanner | Phoenix V|tome|x M300 |
Detector Size (Pixel) | 1996x2000 |
Projection Count | 2500 |
Projections Size | 20 GB |
Output Voxels | 2176x2080x1632 |
Output Vol. Size | 15 GB |
RAM Usage | 50 GB |
For more information on the hardware requirements for the Voxray QRT software, have a look at our hardware guide.