Benchmarking NVENC video transcoding on the Pi
2 days ago
- #Nvidia GPU
- #Video Transcoding
- #Raspberry Pi
- Nvidia GPUs can now run on Raspberry Pi, enabling GPU-accelerated video transcoding.
- Older Nvidia cards (e.g., 3060) can be repurposed for tasks like video conversion or running Jellyfin for streaming.
- Proryanator's encoder-benchmark tool was used to test GPU performance on a Raspberry Pi with an Nvidia 4070 Ti.
- Benchmark results showed significant performance differences between the Pi (limited by PCIe bandwidth) and a high-end PC.
- The Pi's PCIe Gen 3 x1 lane bottlenecked data transfer, affecting NVENC performance, especially for 4K video.
- On a high-end PC with PCIe Gen 4 x16, the 4070 Ti achieved much higher throughput and faster transcoding speeds.
- Jellyfin worked well on the Pi for streaming multiple video streams, despite hardware limitations.
- Power consumption for the Pi setup ranged from 29W (idle) to 130W (transcoding two streams).
- The setup is viable for home use, though performance may be limited by the Pi's hardware constraints.