Guidance to replace my raspberry pi 4

[rsmartapps]

Is this pc capable or good fit for jellyfin for:
Currently I direct play 4k content from jellyfin on a raspberry pi 4.
The pi also has [Stuff we don't talk about here - censored by TDP], vpn, traefik and some personal containers.

Planing to run nextcloud also.

My streaming would hit 2 clients mainly I could reach up to 3. I mainly play 4k content.



Specs:
Intel Core i5 8500
32GB(2x16GB) DDR4 2666MHz
SSD 120GB Kingston A400
HDD 1TB Toshiba
Caja/Placa Base/Fuente Fujitsu D538
Price 150€

My experience with the pi is not bad, but when I move backward or forward it struggle a bit, some subtitles are not been loaded properly.
The drive sometimes fails. I have the feeling that moving to a more reliable hardware would improve my experience a lot.

Having said that I have read so.e threads and they recommend 12th gen cpu I am not Planing to use a dedicated gpu since everything I stream is meant to be direct play. But copy sync would handle transcode.

[TheDreadPirate]

As long as your clients can definitely play your HDR content, the specs for that system are fine. Once you start requiring tone mapping, that MIGHT be able to handle 2 tone mapped transcodes. For regular transcodes, it should be able to handle at least 6 simultaneous transcodes.

[Host-in-the-Shell]

Considering SBCs are not recommended, which includes Raspberry Pis, I think anything that exceeds their computational power will be a significant improvement.

For a server without a GPU, here are the recommended specs:

• CPU: Intel Core i3-7100, Apple M series or newer (excluding Intel J/M/N/Y series up to 11th gen)
  
• RAM: 8GB System RAM (Consider adding more on Windows 11)
  
• Graphics: Intel HD 630, Apple M series or newer
  

I'm assuming the 120GB SSD will be used as a host for the OS. If you don't plan on transcoding multiple streams often then there's no issue; otherwise you probably could use a larger SSD; if you can make it an NVMe, which the Fujitsu D538 seems to support, even better. While storing transcoded files in the HDD is possible, a faster disk is better.

Optional: If you're not on a tight budget, I'd take a look at a mini pc with a 12th gen N100; it may cost around 50€ more, give or take, but it would be a significant upgrade with much better hardware, with low power that could cut costs in the long run, and it would even handle transcoding very well for a media server. The tradeoff is that you won't have the 1TB HDD and you can't really fit one of those inside, but you could later invest on an external disk to attach to it.

All that said, since your goal is mostly to direct play, your current specs are totally fine.

[rsmartapps]

Much appreciated for the quick response. I am a bit concern on the tone mapping, I don't know why would be it necessary what circumstances would lead to and what hardware resources imply to.

What it comes out as for my understanding, the computer I shared would be capable of streaming 4k direct play with tone mapping with a proper ssd drive.

I am not really concern about the money but I don't want to overspend.

Really appreciate your time

[TheDreadPirate]

Tone mapping happens when you play HDR media but the client can't direct play the video for some reason. Right now, if the video needs to be transcoded for any reason, it will tone map HDR to SDR.

To be clear, if the video is already SDR then tone mapping does not happen.

Whether or not something direct plays (no transcoding) is entirely up to the client device's compatibility and not anything to do with the server.

[rsmartapps]

I sold my RPI.

For now I am using my Personal PC, since I am not using it,

I installed Ubuntu Desktop, Added Docker and waken up all containers. I configured QSV despite I have a dedicated GPU, to test performance the CPU is an Intel 9600k.

As I mentioned, I mainly play 4k content, directly is strange I have to perform any transcode, maybe to burn subtitles or audio transcode.

Thanks for sharing your thoughts about and share your knowledge. I will share my experience.

I didn't make my mind about what device to buy, mini, old or beefy.

[rsmartapps]

have run some tests with the i5 9600k and the outcome of them are:
Logs for the ffmpeg command

Code:

Stream mapping:
  Stream #0:0 (hevc) -> setparams:default (graph 0)
  Stream #0:4 (pgssub) -> scale:default (graph 0)
  overlay_qsv:default (graph 0) -> Stream #0:0 (h264_qsv)
  Stream #0:1 -> #0:1 (truehd (native) -> aac (libfdk_aac))
Press [q] to stop, [?] for help
[hevc @ 0x5e54d1fb3a00] First slice in a frame missing.
    Last message repeated 6 times
[hevc @ 0x5e54d4186a80] First slice in a frame missing.
    Last message repeated 6 times
[matroska,webm @ 0x5e54d1023e80] sub2video: using 3840x2160 canvas
libva info: VA-API version 1.21.0
libva info: Trying to open /usr/lib/jellyfin-ffmpeg/lib/dri/iHD_drv_video.so
libva info: Found init function __vaDriverInit_1_21
libva info: va_openDriver() returns 0
[hls @ 0x5e54d1f7b000] Opening '/config/transcodes/d7693e97cf190e14c49d75c93d079a3d-1.mp4' for writing
Output #0, hls, to '/config/transcodes/d7693e97cf190e14c49d75c93d079a3d.m3u8':
  Metadata:
    encoder        : Lavf60.3.100
  Stream #0:0: Video: h264, qsv(tv, bt709, progressive), 3840x2160 [SAR 1:1 DAR 16:9], q=2-31, 122467 kb/s, 23.98 fps, 24k tbn
    Metadata:
      encoder        : Lavc60.3.100 h264_qsv
    Side data:
      cpb: bitrate max/min/avg: 122467815/0/122467815 buffer size: 244935630 vbv_delay: N/A
  Stream #0:1: Audio: aac, 48000 Hz, stereo, s16, 256 kb/s (default)
    Metadata:
      encoder        : Lavc60.3.100 libfdk_aac

the igpu usage running the command was around 75%  the Render and video engines

Code:

Intel Coffelake (Gen9)
1060MHz
11% RC6
8.3/24.5 W
650 irqs/s
IMC reads 8200 MiB/s
IMC writes: 3100 MiB/s