March 23, 2018

NVENC shared context usage

Our Live Transcoder has full support for NVidia GPU decoding and encoding. As you could see from our stress-test article and EC2 tests, Nimble works very good with NVENC.

Usually customers perform decoding and encoding of a relatively small number of streams for input and output and it doesn't affect the performance of neither GPU, nor CPU. However there are cases when our customers use full power of hardware acceleration to process dozens of streams. In this case, some additional performance tuning should be done.

Nimble Live Transcoder allows re-using shared context to optimize resource usage. You may enable it by adding nvenc_context_share_enable parameter into Nimble Streamer config (nimble.conf).
nvenc_context_share_enable = true
nvenc_context_share_lock_enable = true
The second parameter above, nvenc_context_share_lock_enable, will prevent the errors related to NVENC decoding.

Nimble config control is described in this article.

Adding those parameter will enable NVENC context share which will increase the performance of Live Transcoder on high load and reduce number of NVENC-related issue.

If you'd like to manually create context cache, you may follow this article.
You may also find useful our Nimble Streamer performance tuning guide as well as Transcoder troubleshooting tips.

Feel free to visit Live Transcoder webpage for other transcoding features description and contact us if you have any question.

Zabbix monitoring of Nimble Streamer allows tracking server status, SRT streams and NVidia GPU status.

Contact us if you have any other questions or issues with Live Transcoder.

Related documentation

Live Transcoder for Nimble StreamerNVidia GPU supportHEVC support in Nimble StreamerLive Streaming features, Using Amazon EC2 for HEVC transcoding, Nimble Streamer performance tuning.

No comments:

Post a Comment

If you face any specific issue or want to ask some question to our team,
PLEASE USE OUR HELPDESK

This will give much faster and precise response.
Thank you.

Note: Only a member of this blog may post a comment.