Open Access
Table 2.
Profiling results of the TensorRT engine on the RTX 4090 GPU in MagAO-X. The time includes memory transfer to and from the GPU and the inference with the model.
| Inference time | Single precision | Half precision |
|---|---|---|
| Median | 240 μs | 100 μs |
| Min | 235 μs | 96 μs |
| Max | 247 μs | 125 μs |
| 95% percentile | 243 μs | 104 μs |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.