Transcoder GPU Nvidia

[Carmine_Gregorio]

Hi Guys I’m new to Wowza,
I am thinking of purchasing two software licenses for Streaming Engine, but I have a problem with transcoding,
On the server there is a nvidia m10, as you can see by the logs

2021-05-26 12:20:43 CEST comment server INFO 200 - JNI:TranscoderSession.isCUDAAvailable[defaultVHost:prova/definst/stileTV.stream]:NVidia hardware acceleration is available: NVENC9[H264], NVCUVID, CUDA[decode] - - - 123.025 - - - - - - - - - - - - - - - - - - - - - - - - -
2021-05-26 12:20:43 CEST comment server INFO 200 - JNI:NVidiaCUDAUtils.logCUDAGPUInfo[defaultVHost:prova/definst/stileTV.stream]: GPU[0]: name:“GRID M10-1B” computeCapability:5.0 cudaVersion:11020 driverVersion:460.73 memory:1024MB processors:5 cores:640 clockRate:1032MHz - - - 123.025 - - - - - - - - - - - - - - - - - - - - - - - - -

my problem is that i can’t reproduce the local (Transcoder)'s incoming stream
in the access log there is this line
2021-05-26 14:59:58 CEST comment server WARN 200 - JNI:VideoDecoderGenericNVCUVID.initCuda[defaultVHost:prova/definst/stileTV.stream]: cuStreamCreate[m_ReadbackSID]: Failure result:201

indeed in the error log there is this line

prova/definst/stileTV.stream: decodeIterations:32190]: Video decoder pending frame count[150] is greater than limit[150], resetting
WARN server comment 2021-05-26 15:00:52 - - - - - 9731.469 - - - - - - - - JNI:VideoDecoderGenericNVCUVID.initCuda[defaultVHost:prova/definst/stileTV.stream]: cuStreamCreate[m_ReadbackSID]: Failure result:201

I don’t understand where the problem is can someone help me

thanks in advance

[video_streaming]

Can you share what version of Steaming Engine you are currently using for your testing? This could involve several different variables raging from Engine version to Java to whether or not you are using CUDA 10 and the recommended GPU cards.

For example:

• The NVIDIA microarchitecture of your hardware and the NVIDIA drivers you use must support CUDA 10. See Wowza Streaming Engine does not support CUDA 11 (NVIDIA drivers 450.00 and later) for more information and refer to the NVIDIA documentation for installation instructions.
• Wowza Streaming Engine 4.8.8.01 deprecated support for using NVIDIA Kepler GPUs, which will be unsupported in future releases.

Just so you know, this can indeed become a bit overwhelming even for more advanced Wowza users, so even if you’re on a trial, you can still submit a support ticket and have a Wowza engineer work directly with you to troubleshoot this and get you all set up properly.

In the meantime, here is the doc with a list of all related docs for CUDA and NVIDIA including our server specifications for accleration.

https://www.wowza.com/docs/server-specifications-for-nvidia-acceleration-with-wowza-transcoder

[Rose_Power-Wowza_Com]

Update: With Wowza Streaming Engine 4.8.14, we’ve upgraded CUDA support to CUDA 11 for NVIDIA GPU-accelerated live stream transcoding. This upgrade supports using the latest NVIDIA drivers for transcoding at greater scale and speed. The NVIDIA microarchitecture of your hardware must support CUDA 11, and using NVIDIA driver version 460.00 or later is required.

With this upgrade, using Kepler GPUs and CUDA decoding are not supported. For NVIDIA accelerated decoding, use the NVCUVID (also known as NVDEC) implementation instead. For more information about NVIDIA hardware acceleration, see the following the article:

• Server specifications for NVIDIA acceleration with Wowza Streaming Engine transcoding