Unsure about Hardware for Media Server

[JonnyBrate]

Hello everyone,

I am very new to this and very nervous.  
I am looking to get into using Jellyfin and building a home media server to go with it. 
I will not do any computing tasks with it, just for media.

I want to spend as little as possible on this. 
In my house we have about 5 TVs that would want to stream, maybe not always at the same time , but it would help knowing that the server could handle it. It will be running Windows and people will be streaming 1080p to 1080p max. if transcoding is needed. 


My current list of Hardware is as follows:

CPU - Ryzen 5 5600G
MB - Gigabyte A520I
Ram - 16GBs 3200 Gskill
PSU - Seasonic Focus 550w
Some noname SSD


Should I be looking into getting a dedicated GPU or is my Setup OK for my use case as is? Would you change anything?
I want to keep power consumption and price as low as possible while still getting good streams for everyone in my house.

Thank you kindly in advance for your answers!

[bitmap]

No need to be nervous! Jumping into a new hobby should be fun. It can be frustrating at times, but that's part of the learning process -- learning a new thing is almost never easy.

As for hardware, the mantra is: match your media to your clients. In other words, make sure your clients are capable of playing everything you have on your server and you could basically run Jellyfin on a potato and not have to worry about it (seriously, a Raspberry Pi is enough if you follow this mantra).

Most folks don't get into this game before having a media collection, so that's not always plausible and most of the time it's in various formats with different codecs, etc... So let's start with a couple easy questions rather than make you more nervous.

• What kind of clients will your server be streaming to (e.g., Roku, Chromecast, AppleTV)?
  
• What do you mean by "1080p to 1080p max"?
  
• Do you use any virtualization like VMs, Kubernetes, Podman, Docker?
  
• Any network file systems or network-attached storage?
  
• Any other special requirements (e.g., 4K media, using AV1 codec, entire library is in HEVC/x265, everything has 7.1 Dolby Master Audio)?
  

Just from the surface, your use case is pretty simple. If your media library is mostly compatible with your clients (what you will be playing the media on), I don't see why your hardware would present a problem. You could consider bumping up the RAM, but it's not necessary. I don't think the Ryzen processors support any hardware accelerated transcoding, so likely wouldn't cope well with multiple streams, but I don't have much experience with CPU-only transcoding on AMD platforms.

If you really wanted to make 100% sure you could handle whatever you throw at your sever, you could get a basic GPU (you'd be looking for something like an NVIDIA 1060 or better to handle HEVC decoding). A lot of the cards in the cheaper range you'll probably only find used. Another option is an Intel ARC A380, which will do AV1 decoding as well and will run you $150. This isn't really necessary at all if you make sure that your media matches what your clients can consume, but you'll always find an edge case; it's up to you whether you want to prep for transcoding or just let the clients that can't play back the media deal with it.

[Efficient_Good_5784]

My Jellyfin server has the 5600G CPU as well. I have HWA set with VAAPI.

Using the iGPU, it can transcode a single HEVC 1080P to AVC 1080p stream at around 120fps-170fps.

Worst case scenario for you would be everyone transcoding at once. As long as the total framerate of each stream (added together) falls bellow the max fps the CPU can transcode and pump out, then there won't be any buffering.

The rest of your specs seem fine. If you find out that the iGPU of your cpu can't handle your daily workload, you can always buy a GPU in the future to handle transcoding.

[JonnyBrate]

• What kind of clients will your server be streaming to (e.g., Roku, Chromecast, AppleTV)?
  
• What do you mean by "1080p to 1080p max"?
  
• Do you use any virtualization like VMs, Kubernetes, Podman, Docker?
  
• Any network file systems or network-attached storage?
  
• Any other special requirements (e.g., 4K media, using AV1 codec, entire library is in HEVC/x265, everything has 7.1 Dolby Master Audio)?
  

Hello and thank you for your in-depth follow up questions.

1. The server will be streaming to four Samsung Smart TVs from one to four years old using the jellyfin app, which will be deployed on them. One TV will be using a chromecast.
   
2. I read about transcoding (still completely unsure if it will be the case since I don't understand it 100% since English is not my mother's tongue and more complex topics take me a while to fully grasp) and that it uses quite some power.
   
3. No virtualisation is planned.
   
4. The server will be using 2 Seagate Ironwolf 12TB HDDs in RAID (don't know which one yet)
   
5. 1080p will bew the goal since my Kids and in-laws don't care about 4k. I also plan on keeping with x264.
   

Thank you in advance for any further help!

---

My Jellyfin server has the 5600G CPU as well. I have HWA set with VAAPI.

Using the iGPU, it can transcode a single HEVC 1080P to AVC 1080p stream at around 120fps-170fps.

Worst case scenario for you would be everyone transcoding at once. As long as the total framerate of each stream (added together) falls bellow the max fps the CPU can transcode and pump out, then there won't be any buffering.

The rest of your specs seem fine. If you find out that the iGPU of your cpu can't handle your daily workload, you can always buy a GPU in the future to handle transcoding.

Thanks for the reply!

How many users do you have in your network? Does that make any difference?

Thank you in advance!

[Efficient_Good_5784]

2. I read about transcoding (still completely unsure if it will be the case since I don't understand it 100% since English is not my mother's tongue and more complex topics take me a while to fully grasp) and that it uses quite some power.

Think about transcoding this way:

Lets say that you only know how to read English. You can natively read (direct play) all books in English.
You get a book that is written in Italian. You have no idea how to read it, so you give up and don't attempt to read the book.

If you do want to read it, you now have two options:

1. Grab an Italian to English dictionary. You are able to translate the words using the dictionary, but it takes you a long time and a lot of effort to read through the book.
   
2. Instead, you can get someone that knows both English and Italian to translate and rewrite the book into English. Now, you can read the rewritten book with minimal effort.
   

You can think of option 1 as software transcoding, and option 2 as hardware transcoding.

This same concept applies here. If the client (Jellyfin app on your TVs in this case) doesn't know how to natively read the media, your server will attempt to transcode the stream so that the client can natively read it.

You got two options to transcode. Software (CPU) or hardware (GPU) transcoding. GPU transcoding is also referred to as "hardware acceleration" or HWA. CPU transcoding usually is slow and takes up more power. GPU transcoding though is usually faster and requires less power. With CPU transcoding, your CPU is busy which means everything else that needs the CPU will have to wait a bit longer. With GPU transcoding, most of the workload is taken off of the CPU cores so that the CPU can handle other things at once.

Jellyfin lets you to disable transcoding completely, or have it enabled with either CPU or GPU transcoding.

How many users do you have in your network? Does that make any difference?

I'm the only one that uses my server. But I have tested how many streams I can open in a browser before. I don't remember exactly, but 5 streams were not much of an issue with the 5600G doing HWA.
I will say that using the Android TV Jellyfin app, the only times I have had issues are with subtitles sometimes causing media to not play.

[JonnyBrate]

2. I read about transcoding (still completely unsure if it will be the case since I don't understand it 100% since English is not my mother's tongue and more complex topics take me a while to fully grasp) and that it uses quite some power.

Think about transcoding this way:

Lets say that you only know how to read English. You can natively read (direct play) all books in English.
You get a book that is written in Italian. You have no idea how to read it, so you give up and don't attempt to read the book.

If you do want to read it, you now have two options:

1. Grab an Italian to English dictionary. You are able to translate the words using the dictionary, but it takes you a long time and a lot of effort to read through the book.
   
2. Instead, you can get someone that knows both English and Italian to translate and rewrite the book into English. Now, you can read the rewritten book with minimal effort.
   

You can think of option 1 as software transcoding, and option 2 as hardware transcoding.

This same concept applies here. If the client (Jellyfin app on your TVs in this case) doesn't know how to natively read the media, your server will attempt to transcode the stream so that the client can natively read it..

Thank you that explanation quite did it for me. Might want to look into getting a cheap GPU. I just may have issues with he power draw. I want to keep it as low as possible. Any Ideas on that maybe?

Thanks in advance!

[Efficient_Good_5784]

Just make sure to use HWA. My 5600G uses around an extra 30 watts of power when HW transcoding.
I don't remember exactly, but with software transcoding, it uses around an extra ~50 watts of power.

If you're curious about your server's power consumption, buy a watt meter off of the internet so that you can see how much your entire server uses when idle and when transcoding.

I would say you don't need to buy a GPU. Go ahead and set Jellyfin up with your current equipment and see how it works.
You would only need to buy a dedicated GPU if the CPU's iGPU doesn't handle your workload or some new file type doesn't work well with the iGPU.

Just so you know, in terms of power consumption, a dedicated GPU will still use power even when it's not being used. It's not a lot of power, but it's still a few watts which adds up over long periods of time.

[bitmap]

Thanks for jumping in! I think you've got it covered and answered the questions nicely. Great explanation of transcoding!

[JonnyBrate]

Just make sure to use HWA. My 5600G uses around an extra 30 watts of power when HW transcoding.
I don't remember exactly, but with software transcoding, it uses around an extra ~50 watts of power.

If you're curious about your server's power consumption, buy a watt meter off of the internet so that you can see how much your entire server uses when idle and when transcoding.

I would say you don't need to buy a GPU. Go ahead and set Jellyfin up with your current equipment and see how it works.
You would only need to buy a dedicated GPU if the CPU's iGPU doesn't handle your workload or some new file type doesn't work well with the iGPU.

Just so you know, in terms of power consumption, a dedicated GPU will still use power even when it's not being used. It's not a lot of power, but it's still a few watts which adds up over long periods of time.

Right. I will set up and see if I need a dedicated GPU and might get one down the line. Thanks again!