Hey, it seems, if h264 input video contains frames with slices, there is no difference in decoding latency, nor throughput for nvdec when compared to video with non-sliced frames. How is it possible? Software decoders seem to heavily gain from sliced frames (if sliced-based threading is enabled).
Slice threading is not the same as slices of H.264. Slice threading is a technick of multithreading. Most files have 0 slices anyway.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| NVdec error handling with multi slice frames | 3 | 65 | April 2, 2025 | |
| Zero latency possible or not? | 2 | 2026 | October 12, 2021 | |
| Slices lost to packet loss (h264) | 0 | 433 | July 9, 2020 | |
| Splitting videos to frames and hardware decoding | 1 | 652 | August 16, 2016 | |
| Slice encode/decode support | 6 | 2359 | October 18, 2021 | |
| Is there any specific optimization in NVDEC? | 1 | 633 | November 28, 2018 | |
| Video tearing when modifying slice-header-spacing in omxh264enc | 5 | 1537 | October 18, 2021 | |
| Multiple slices on Jetson NX Jetpack 4.6 | 11 | 849 | April 5, 2023 | |
| Decode h.264 video and convert to cv::Mat | 0 | 840 | November 13, 2018 | |
| Jetson Video Encoder: problems with GDR, slices and Slice Level Encode | 8 | 1462 | December 7, 2022 |