It’s clearly less than ideal, but better than not being able to play something at all. Make it an opt in, user selectable feature and there won’t be any issues. This will be less of a problem (unless messing with UHD content) as more people have 802.11ac networks. I’m not particularly worried about bandwidth at this point for my specific use case.
@justflie said:
It’s clearly less than ideal, but better than not being able to play something at all. Make it an opt in, user selectable feature and there won’t be any issues. This will be less of a problem (unless messing with UHD content) as more people have 802.11ac networks. I’m not particularly worried about bandwidth at this point for my specific use case.
With network bandwith usage, I was referring to mobile networks with metered connections.
Ah. And I was thinking about local networks only. Got it. Clarified!
Sent from my iPad
The idea is to have a lower powered machine manage all transactions and then offload to a more powerful machine to handle transcoding within the local network. Ideally, these machines would be connected by serial or on a separate subnet. Each machine would need to be running its own PMS.
It would not need to run a PMS, but at least a listener that connected back to your PMS, and then the transcoder parts. All it would need is access to the media, and a way to pipe that back through the PMS.
1 Like
Plex would need to allow multiple ports to be accessed for internal traffic for transcoding. IMO, I would option PMS only on specific machines that can handle transcoding. Plus it adds a layer of redundancy.
I see an issue of latency, but it may only be on initially starting media.
That is not necessarily true, it would only need one port open with transcoder identifiers. so the main PMS would know who is who. It would probably do well to be able to communicate a CPU/GPU load for each, then it can load balance between them all. Or, it can send all transcodes to ALL boxes and just have them all work as a CPU/GPU farm as render farms do. I think it would probably be easier to do it a single xcode per box, but I won’t complain if they want to do it farm style.
I think the “load balancer-like” approach would be the best - i.e. one machine exposed to the internet (just like now with a single PMS) with the traffic flowing through the main machine to transcoders and back, and several machines behind it, acting as transcoders.
PS: An ability to automatically wake up and send to sleep every transcoder machine would be nice 
. I would sacrifice some delay in starting the stream just to have something like that 
.
Would love this feature - saves the pain of trying to move the PMS and a simple inexpensive lower powered device like a shield would be perfect.
The better solution could be having another powerful PMS but using the database from the initial PMS ( for example from low powered NAS ).
Of course you could build another PMS with same folders ( using network shares ). But when you watch something from the powerful PMS, it wouldn’t update on the other one.
What I mean is like this.
At home I have a Atom C2751 NAS with 50 TB of storage. It is powerful enough to manage downloads and playback 1080p x264 movies etc… But when watching HEVC encoded 1080p or 2160p files I see that quality is going down to playback the file smoothly ( I’m watching on AppleTV + projector and it almost always transcodes the files )…
I also have my desktop PC which has a poweful CPU+GPU ( i7 6700K+GTX 1080Ti )… I installed PMS on that one also, and used same folders for libraries. It’s almost exact copy of the other server. But the problem is what I watch is not mirrored on the other one obviously. If there was a solution for that it would be perfect ( syncing watch status on two PMS servers with same movie/tv shows )… That would be of course better solution, but it may be harder to do then what I initially suggested.
So, please give us capability to use the database from another Plex media server, so we can use the powerful server when watching heavy duty content…
This was requested back in 2014.
It is 2018. Unfortunately, as much as I hate to say it. Plex developers will never, ever implement this feature. If they haven’t implemented it in 4 years, I doubt it will happen ever. Which is a shame. Because I would love something like this.
What do you expect? They don’t listen to us at all. There are multiple feature requests that haven’t been implemented and instead we get photo tagging and severely restricted Plex Cloud. They just ignore our input and do what they think we need - which is usually wrong.
Plex has not really shown a concern with fixing broken things or releasing features which would not net them new paying members unfortunately. I mean, base functionality like subtitles STILL do not work properly after being “fixed” 4 times now. 4k streaming on CCU is still only partially functional. We don’t even have CCU profiles yet over a 1.5 years after release! Hardware xcoding for the masses, which has been partially delivered is the reason why we continue to fight to get this feature, as they listened and kinda sorta delivered that.
With 4k picking up BIG in 2018, 4k tvs now being cheap, this is going to be a MAJOR pain point for… well… everyone soon, and I think we will see some very large fights on who will be #1 media center soon on whoever is going to solve this issue first. With it taking a 4000-8000 passmark score per 4k xcode, The only two viable solutions are distributive xcoding, or someone coming out with dedicated GPU hardware which will allow multiple xcodes without the quality loss that currently exists. Out of those two I know which is easier and cheaper to R&D… but the question is which will these companies attempt to actually do. They cannot ignore this too much longer, 8k is on the horizon already, and Plex is already falling behind other media centers out there on 4k compatibility alone, much less innovating. Other things which are being ignored are newer codecs like TrueHD which force a transcode regardless of support on your hardware.
Either way, traction is GOING to have to be made, and soon… so what’s it going to be Plex?
Distributed encoding probably won’t ever come to pass in any official way from Plex. This is just me personally speculating on this and nothing more.
Plex has and is coming up with several different ways to allow us much more “processing” power for transcoding. We now have official QSV/GPU support which allows offloading transcoding to hardware and it works very well especially on the latest generation hardware.
The new TYPE FIRST menu systems out on PMP and XBOX and due to arrive on other clients also can help in this regard as well. With TYPE FIRST the user doesn’t have to select a server but now selects a Library. The library can be on any server you run connected to your account that you share with the user. This can allow you to come up with some unique setups to help distribute the load. For example:
Plex1: Movies Archives
Plex2: TV Shows Archives
Plex3: Sports and Educational libs
Plex4: Music and Photos
Plex5: New Releases
Plex6: DVR Recordings
These can be combinations of NAS boxes (music and photos) to ShieldTV Servers (DVR) to PCs running Plex Servers. Granted this isn’t the same as distributed transcoding but it does allow you to break up your libraries and place them on different hardware and therefore gaining additional “native” transcoding power via the use of multiple machines (with or without QSV/GPU support). To the user it’s just one big Plex server.
Of course with the above setup you could have 2 Movie servers and 2 TV Show servers and you manually split the users so half use one server and the other half use the other server while they all use the same Plex3 to Plex 6 servers. TYPE FIRST while not a “clean” solution does gretly expand the size of the system you can run for family and friends!
Carlo
Not what we want, but that is certainly a step in the right direction. Won’t solve the issues at all with 4k and 8k streams, and QSV/GPU is only VERY partially supported and the quality is highly degregated at the moment with blocky edges. In the case of 4k and even 8k kinda defeats the purpose 
Still, that IS a step in the right direction, and there is no way Plex as a company does not know that they WILL have to come up with a real solution, and soon.
I’ll disagree with quality. I find with recent CPUs or a good AMD video card the quality is better than you can achieve in real time using the CPU. Stop reading web sites who compare multi hour conversions using CPU to HW as this isn’t the same as real time transcoding. For real time, HW will/can easily produce better.
I also haven’t seen any sites that have done recent comparisons using the latest hardware nor anything that would be similar to how Plex uses transcoding in real time.
But yes, it’s not the same thing as what’s being asked for here. But as I said, I doubt it will ever happen. Maybe 2 years ago but not now due to the way audio has to be processed to stay within the legal Dolby requirements. Remember Plex is a licensed Dolby customer so it can’t just do things any old way as they have to abide by their license. This makes distributed transcoding that much more difficult to achieve.
At this point in time it’s fine to ask for things like this but best to start thinking of ways to expand your system using the tools and methods available to us now.
Just my 2 cents,
Carlo
PS I myself would LOVE to have distributed encoding especially if we could also use ShieldTV devices which can already do HW transcoding and can run the Plex Media Server. It would be great to have “cheap” clients that could also transcode for the server as well!
@cayars said:
Distributed encoding probably won’t ever come to pass in any official way from Plex. This is just me personally speculating on this and nothing more.Plex has and is coming up with several different ways to allow us much more “processing” power for transcoding. We now have official QSV/GPU support which allows offloading transcoding to hardware and it works very well especially on the latest generation hardware.
The new TYPE FIRST menu systems out on PMP and XBOX and due to arrive on other clients also can help in this regard as well. With TYPE FIRST the user doesn’t have to select a server but now selects a Library. The library can be on any server you run connected to your account that you share with the user. This can allow you to come up with some unique setups to help distribute the load. For example:
Plex1: Movies Archives
Plex2: TV Shows Archives
Plex3: Sports and Educational libs
Plex4: Music and Photos
Plex5: New Releases
Plex6: DVR RecordingsThese can be combinations of NAS boxes (music and photos) to ShieldTV Servers (DVR) to PCs running Plex Servers. Granted this isn’t the same as distributed transcoding but it does allow you to break up your libraries and place them on different hardware and therefore gaining additional “native” transcoding power via the use of multiple machines (with or without QSV/GPU support). To the user it’s just one big Plex server.
Of course with the above setup you could have 2 Movie servers and 2 TV Show servers and you manually split the users so half use one server and the other half use the other server while they all use the same Plex3 to Plex 6 servers. TYPE FIRST while not a “clean” solution does gretly expand the size of the system you can run for family and friends!
Carlo
Soo, the server selection was just moved down one level. That’s not changing anything in my opinion ( and the old way is still present in other players).
It should be like it is with “Play version” option. If a certain item was present in multiple servers, the user could choose which server to use or it could be forced upon.
@cayars said:
I’ll disagree with quality. I find with recent CPUs or a good AMD video card the quality is better than you can achieve in real time using the CPU. Stop reading web sites who compare multi hour conversions using CPU to HW as this isn’t the same as real time transcoding. For real time, HW will/can easily produce better.I also haven’t seen any sites that have done recent comparisons using the latest hardware nor anything that would be similar to how Plex uses transcoding in real time.
But yes, it’s not the same thing as what’s being asked for here. But as I said, I doubt it will ever happen. Maybe 2 years ago but not now due to the way audio has to be processed to stay within the legal Dolby requirements. Remember Plex is a licensed Dolby customer so it can’t just do things any old way as they have to abide by their license. This makes distributed transcoding that much more difficult to achieve.
At this point in time it’s fine to ask for things like this but best to start thinking of ways to expand your system using the tools and methods available to us now.
Just my 2 cents,
CarloPS I myself would LOVE to have distributed encoding especially if we could also use ShieldTV devices which can already do HW transcoding and can run the Plex Media Server. It would be great to have “cheap” clients that could also transcode for the server as well!
But the problem is, Quick Sync is not really that powerful, and Nvidia GTX and lower level Quadro cards are limited to 2 streams per system. Does AMD impose such artificial limit?
@Monsters_Grin said:
Soo, the server selection was just moved down one level. That’s not changing anything in my opinion ( and the old way is still present in other players).
It should be like it is with “Play version” option. If a certain item was present in multiple servers, the user could choose which server to use or it could be forced upon.
No, the “server selection” is eliminated more or less. It’s more a concept of being able to play a file from a library that can exist on any server.
But the problem is, Quick Sync is not really that powerful, and Nvidia GTX and lower level Quadro cards are limited to 2 streams per system. Does AMD impose such artificial limit?
I would completely disagree with you on this. QSV or the correct GPU will run circles around what you can do with a CPU for transcoding in REAL TIME. Try transcoding 10 bit HEVC to H.264 in real time. Now do 4 to 6 of them at the same time. Good luck with that with CPU only but with HW it’s just as easy to transcode one stream as it is to transcode 4 streams (assuming your HW is up to it) and the result will look just as good with 1 stream vs 4. With CPU it’s impossible to real-time encode 1 stream with results that are as good as modern QSV let alone trying to do multiple which WILL degrade as the quality has to be reduced to give the CPU cores the chance to do real-time encoding.
Nvidia consumer cards are limited to 2 transcodes. There are no limits for QSV or AMD GPUs. However for now Plex can only use one HW device. I’m sure in time Plex will allow multiple HW GPU/devices to be used in the server which will allow for very powerful Plex servers. ← this will be much more powerful for the amount of dev work needed.
@cayars said:
No, the “server selection” is eliminated more or less. It’s more a concept of being able to play a file from a library that can exist on any server.
When I go to “Shows” in PMP, it lets me choose which server’s “show library” I want to browse. If it was done like “Play version” function, content from all servers the account has access to could be shown as one big library with source selection based on user’s preferences (quality, server order maybe, connection speed).
@cayars said:
I would completely disagree with you on this. QSV or the correct GPU will run circles around what you can do with a CPU for transcoding in REAL TIME. Try transcoding 10 bit HEVC to H.264 in real time. Now do 4 to 6 of them at the same time. Good luck with that with CPU only but with HW it’s just as easy to transcode one stream as it is to transcode 4 streams (assuming your HW is up to it) and the result will look just as good with 1 stream vs 4. With CPU it’s impossible to real-time encode 1 stream with results that are as good as modern QSV let alone trying to do multiple which WILL degrade as the quality has to be reduced to give the CPU cores the chance to do real-time encoding.Nvidia consumer cards are limited to 2 transcodes. There are no limits for QSV or AMD GPUs. However for now Plex can only use one HW device. I’m sure in time Plex will allow multiple HW GPU/devices to be used in the server which will allow for very powerful Plex servers. ← this will be much more powerful for the amount of dev work needed.
I don’t argue the quality - it’s pretty good on QSV. But I’m curious on how different it’s performance is between processors. Do you have that kind of data? I can’t find it.
The ability to use multiple devidces would be nice to have, but it would be unfortunately useless on nVidia hardware. Their limit is not per GPU, but per system. Don’t know what to do with QSV though - Unless they release some kind of add-in cards, like those encryption accelerators that use Intel Quick Assist Technology - see HERE . . .
@Monsters_Grin said:
@cayars said:
No, the “server selection” is eliminated more or less. It’s more a concept of being able to play a file from a library that can exist on any server.When I go to “Shows” in PMP, it lets me choose which server’s “show library” I want to browse. If it was done like “Play version” function, content from all servers the account has access to could be shown as one big library with source selection based on user’s preferences (quality, server order maybe, connection speed).
Currently PRT pretty much works like this (3rd party distributive). You select a video to play, then the PMS dings all your xcode boxes (ssh session and uses that ssh to xfer too) and finds out which one is the least busy, and starts the xcode on that box. It mostly works right now, and if Plex wants to use that setup for their own, I am betting the guy that made it would be happy to oblige.
1 Like
