Do I hear any volunteers to port all the Intel and nVidia codecs and libraries to native AMD and be operable on all supported OS platforms ?
so I own a TVS-873 and just installed a NVIDIA 1050 TI card in it. I had to use a 16x to 8x cable to get it connected. The QNAP sees it, but container station wants to latch on to it before I can attach it to anything. Are there any steps that I would need to take to get PLEX to use it? I would love to beta test if there is code I can load.
http://philbennas.myqnapcloud.com/NVIDIA-1050ti-on-tvs-873.jpg
http://philbennas.myqnapcloud.com/QNAP-tvs-873%20with%20NIVIA%201050-TI_Instalation.jpg
Not supported.
Nvidia card is ONLY supported in the TS-x73 family and NOT the TVS-x73 family for QTS usage.
So I have to ask, is that because QNAP says so, or because no one has ever tried to support it on the TVS-x73 family? If the array can see it and use it, whats the difference? My implementation may not be conventional or âSupportedâ, but how many of us out there have Frankensteined gear together to do what we needed? I encourage the PLEX developers to look into this implementation and see what it would take to make it work with PLEX.
I have someone checking with QNAP for the exact details, but my understanding is that is a HW limitation due to the TVS-x73 already having an embedded Radeon GPU in the AMD CPU.
The TS-x73 family has a slightly different AMD CPU than the TVS-x73 family which does NOT have the embedded GPU.
For Nvidia based transcoding, Plex relies on QNAPâs Nvidia Driver implementation/support and this is currently NOT a supported use case for that card/driver (the TVS-x73 family).
@drphil02, have you tried creating a PMS in the container station? The new version of the container station (released last week, I think) allows you to add the GPU to a container created from the default Plex docker image. I am very interested in your experiments, as I am on the same boat as you (TVS 873).
@ChuckPa Do you mean writing the code?
Yes, I was referring to all the coding work which has to go to get all the appropriate libraries ported to all the Plex platforms with accommodations for all the different deployment environments.
Everything currently in public domain would need upgraded and âfittedâ.
Itâs not an overly complex task but it daunting and detail rich none the less.
Itâs several dedicated man-months of work at a standard 40hr week schedule.
Well⊠Than that is obviously something that would take some funding to be achieved in reasonable time, as open source work done in peopleâs free time takes quite a bit longer.
It does take funding but more importantly takes the dedicated eyes and hands to get it done.
The difficulty comes in that there is no direct 1-1 mapping. Itâs not a âreplace A with Bâ type port.
One must understand the subject matter and intended functionality in order to write the appropriate lines of interface code to include with it.
1 Like
That is true, but again right now limited (to my knowledge) to the Nvidia based GPU cards.
Qnap was already doing this for the QUAI specific applications (TensorFlow, etc.) but with the new release just added a GUI option to open the GPU usage up to other Docker use cases.
That didnât really change WHICH GPUâs are supported for that use case (NVidia cards).
Exactly, but in @drphil02 's case, he has a GTX 1050 installed already.
Let me say this one more time please?
I did the development work on the TS-877.
I wrote it such that, when QNAP makes NVIDIA_GPU_DRV (the drivers package) available for others, Plex will see it and use it.
I cannot guarantee how well any particular GPU card will work.
GPU selection and individual card performance is out of scope.
Sorry I just got a notification that the thread was active. In response to @pcamargo, Yes I had tried a docker PMS and it wouldnât use the 1050. I appreciate someone looking into my use case
Have you done that after updating the container station to the latest version as of last week?
He does, but in an unsupported configuration (TVS-x73 system with embedded GPU).
Yes I have. The updated container station exhibits the same behavior
Seems I failed to clarify what I implemented.
I implemented native host app support only. This is because of how QNAP has structured their implementation.
I did not implement Container Station - Docker support.
On QNAP, access to their GPU mechanism would require a QNAP-specific version of Docker.
This is contrary to how Docker is supposed to work. Everyone touts Docker as being universal however hardware passthrough is anything but universal.
Run the native application, put the card in QTS mode, and you will discover nVidia GPU encoding is enabled.
That doesnât really work with QTS 4.3.6, as the available NVIDIA drivers are too old. Or am I missing something?
What doesnât work on 4.3.6?
The native Plex server package (qpkg) works with the current QNAP provided NVIDIA_GPU_DRV package
