Good news selfhosted crowd! I pitched the idea of setting up a server, and she is down with the idea!
I want to run Linux and host things like Jellyfin, Audiobookshelf, Navidrome, maybe Plex (I currently host it off my Mac desktop but would move it to a dedicated server).
I know you can host Plex et al on something like a Raspberry Pi, but I want a bit more power. I want something that can push at least 2-3 4K streams.
I’m considering targeting 12th gen i5 with 16GB DDR4. Maybe 32GB if I can swing it. Is that overkill? I know you want 8th gen Intel for hardware transcoding of either x265/HEVC or AV1 (probably the former as the latter is newer).
There are other projects I’d like to run. I’d like to be able to host my own LLM/AI but I’m not trying to go performance overkill here. This would be my first server.
I am somewhat comfortable using the command line, and I would be remoting into the server via my Macs (I have a desktop and a MacBook). However, Linux can run Windows games via Proton and that’s something I’m interested in for some older titles that aren’t on Xbox (like Deus Ex, which came out in 2000 and should run on integrated graphics). I’m not inexperienced with Linux and have been using it off and on (mostly off unfortunately) for over 30 years. My go-to would be Ubuntu, but I’ve heard great things about KDE lately and I know the Internet is in love with Mint these days. Would the distro really matter all that much as long as I have the required services?
You must log in or # to comment.
I’d always recommend not spending any money until you know what you really want and even then use any spares/old stuff you already have. The first build should be a true “budget build” :D
my i7 3700k with 8gb of ddr3 handles multiple 4k streams just fine. it doesn’t transcode on the fly super well, I transcode everything to hevc before putting it on the server. it will happily transcode a single 4k stream in the rare case a client doesn’t want hevc.
point is, don’t target latest gen simply because it’s latest. 12th gen is definitely newer than I’d opt for, and while 16gb is cool, 32 is absolutely overkill for the use cases you listed (other than the llm, which others properly list as “don’t worry about it unless you have 3k in your back pocket to shell on gpus”)
12th Gen is definitively neat, but overkill for what you are doing. 8th or 9th gen is fine. 16gigs is fine, 8gig would also work
I really like the Lenovo m720q line of mini pcs (used/refurb) but they will not fit an hdd.
/edit: I mean an 3,5” HDD. They will fit one 2,5” HDD fine.
Huh? I have the m93p, 710q and P330…and they can all mount internal 2.5" SATA drives (with the latter 2 being able to take m.2 also). The 720 must be the odd duck out.
No you are right, the m720q will fit an 2.5” and also m.2 drives.
Sorry, what I meant is that they will not fit 3,5” hdds, let alone multiple ones.
Ah, completely understandable. Yes, NONE of them fit the standard 3.5" hdds.
I really like the Lenovo minis. They’re great little machines.
Just one thing that could be easy to forget. Power usage. Cost adds up as the hours and days pass :)
Acronyms, initialisms, abbreviations, contractions, and other phrases which expand to something larger, that I’ve seen in this thread:
Fewer Letters More Letters AP WiFi Access Point IMAP Internet Message Access Protocol for email LXC Linux Containers NAS Network-Attached Storage PSU Power Supply Unit Plex Brand of media server package PoE Power over Ethernet SATA Serial AT Attachment interface for mass storage SMTP Simple Mail Transfer Protocol SSH Secure Shell for remote terminal access TLS Transport Layer Security, supersedes SSL VPN Virtual Private Network nginx Popular HTTP server
[Thread #108 for this comm, first seen 21st Sep 2026, 05:00] [FAQ] [Full list] [Contact] [Source code]
For distro I’d recommend God’s Chosen Stock Debian. You want prioritize smooth upgrade paths from one major version to another since the server should last a long ass time, and Debian is that. Debian has the eyes on it to make it chill and if there are things to do they’re caught and noted (and more often something is set up to just do it for you).
So no, distro doesn’t matter as long as you can get your software going. But you’re not going to beat the insane stability of Debian. Any OS based on it is just an echo, which may even involve custom kernels which add more work for the specific distro maintainers and may slow critical security updates.
I run no GUI but you can def have one. I just SSH in. I do everything in containers (Docker 1st and I’m moving to Podman - but Podman is too heavy a lift to start, Docker compose files are always right there, use em) and LXC containers run by Incus (it has a web gui as well!).
I’ve been looking at “games on whales” https://games-on-whales.github.io/ to stream games from a beefy server to weaker computers and tablets. Might be a good route for your games without having to directly game on the server.
I have a 4 core 7th gen Intel CPU and it seems fine with doing x265 transcoding on multiple streams, but I’ve never tried finding out how many 4K ones it can do at once. I have 32 GB of RAM (I bought cheapo laptop RAM before the true AI shitshow, praise be, stuck them in with a DIMM board cooked up by China). I had 16 before that, and it was fine. But I just kept adding services.
An LLM will eat your RAM, you’d want 32 GB to be able to run a 16 GB one. But a GPU will be much, much faster. LLMs are just matrix math - that’s why they’ll never be able to think for real - and GPUs are sicko good at that kinda math (cause screens are a matrix!). I don’t fuck with LLMs though, no ethics in that whole shitshow.
Good luck!
Thanks. The more I see, the more I realise I should just leave the AI stuff out. AI and gaming were outlier possibilities anyway, though I’m way more interested in gaming than AI. Right there with you on AI ethics. I don’t like seeing them steal art and rip off artists, but they can be good at tracking down solutions on various forums. At least, they do it faster. But it sounds like running one might be outside my budget, so I’m more interested in running services, and light/classic gaming, since I have an Xbox (Series X).
I second debian!
I initially had some trouble getting streamed games to use the integrated gpu while the server was headless. I’ve heard you can use a dummy plug into one of the hdmi ports, but I found that using “xrdp” as the remote desktop client works as well. I hope sharing this might save you the hassle evetually.
I third Debian. I was a Ubuntu server fanboy for the longest time. I kept getting scared by “but it uses old packages” message about Debian. But it’s way overblown. Also, if you are using Docker for hosting your services, system packages don’t matter.
Distro kind of matters.
Its a server, stability is the top priority. I’d avoid Ubuntu personally, and recommend Debian and Debian variants. For your first server (and your 4th+), I’d suggest proxmox. Its Debian with virtualization and a web interface, making it easy to manage services.
I’d also mention that you do not want a DE for your server. Its a lot of overhead for no real benefit, instead you should think of it as a headless machine you can remotely access - which is also what will make proxmox handy for you.
Intel wins in transcoding in most scenarios, so an intel-based system with an iGPU is a solid target. Full transcoding of AV1 starts with 14th gen, but that includes to AV1. For decode of AV1 for transcoding, 11th+. For anything else, 7th+.
With ram you can do more with more, but leveraging virt will make the load a lot lighter. Ive got systems with 32GB, but they are also hosting whole ass VMs and not just lxc’s.
Heard good things about proxmox. What do you mean it’s Debian? It’s an OS? I thought it was some kind of VPN thing that gets your services out onto the web so you can access them remotely or something like that.
I’m inclined to agree I don’t need a DE (desktop environment, though I suppose I’m the only one who had to think about that for a second as the new guy) but I was also interested in checking out some Steam games. Would that be something I could spin up separately (Steam via Proton) or would that be better as its own project with its own goals?
14th gen is kind of a price jump, but if I’m only targeting 16GB RAM (per another reply), it might be doable. Having my Mac just convert the AV1 to HEVC (or just avoiding AV1 entirely) sounds like it would make more sense. I’m seeing a lot of refurb deals on 11th gen i5s and that’s what I was thinking.
What are LXCs?
Debian is the OS.
Proxmox builds on Debian to turn it into a hypervisor. That’s an OS for managing VMs and containers. Proxmox (like most hypervisors) gives you a handy webinterface.
There are easier hypervisors to use, but Proxmox is great for learning, and allows you to expand well later.
tbh, except for gaming and AI, you don’t need nearly as much processing power as you might think. I run most of my stuff on a (fairly new, can’t remember which one exactly) i5, and it sits at an average 2-3% CPU usage, with the maximum its recorded being 10.9% (there’s two VMs and 8 containers running on it). 11th gen i5 will probably be more than enough to start with.
In my non-professional opinion, what you really want to look at is how much power your system is going to use when idle, since that’s what (small) servers are doing most of the time. This is not something specified by manufacturers, and it depends on all the parts going into the computer (CPU, mainboard, PSU, fans, how the BIOS is set up, …).
I would also recommend getting at least 16 GB of RAM (honestly I would say 32, if not for the current prices). Containers don’t use too much, but if you want to set up VMs for services, each VM will want at least 2-4 GB for itself.
LXC-Containers (https://en.wikipedia.org/wiki/LXC) are like extremely lightweight and slightly less secure Linux VMs. They are similar to Docker, but while Docker is containerizing applications, LXC containerizes Linux distros.
Yup, proxmox is great.
Note that you can’t (or rather, shouldn’t) run docker containers on proxmox. Most people run a vm and then docker inside that. And yes, I use Debian for it
You can give proxmox a docker container reference, but it’ll download and convert it to LXC, which means you can’t update the container. You’d have to update the container’s os and the app manually, or redeploy the container and reattach the storage.
If I’m running Ubuntu server can I backup my docker files and will Debian find my mdam raid without issue?
For mdadm, it stores the metadata on the drives themselves. As long as your os is separate, yes, it should be just fine and detect automatically. After install it should just be a mdadm --assemble --scan and off to the races.
That said, I will always recommend backups. You never know what will happen.
In terms of docker, yes you can back things up and move them over. That said I recommend having your docker contents that need to be persistent as part of a mount somewhere, so that instead of needing to copy them around, you just have a compose that points to a mount.
Makes backups easier too.
Yeah, I’m thinking of getting a sata ssd and putting it in one of the hot swap bays as I have 6/8 filled with 14tb HDDs is raid 6. Then putting all the docker stuff there so it’s not on the boot nvme.
A solid call either way, easier to manage and safer imo
something that can push at least 2-3 4K streams.
12th gen i5 with 16GB DDR4. Maybe 32GB if I can swing it.
unless you’re transcoding each stream that should work, at least for Plex.
Currently running 10th gen i5 with 32gb of DDR and can run one 4k transcode, five 1080 streams, and four live TV streams. though, the transcode is laggy but if you pause and let it buffer it should work.
keep in mind I’m also running headless transmission, a VPN, tor, and like 8 other services on the same server.
if all you’re streaming is 4k, you will hit a point of diminishing returns on network throughput. I’d recommend getting a switch with SFPs and connect your server up to a 2.5gbe or even a 10gbe (though overkill but the sky is the limit). as long as all your stream boxes are connected through that switch on their regular 1gbe they should be fine.
personally I run a opnsense router/firewall and have everything passed through an unlocked Brocade ICX6610-48P-E. you can find them on eBay. FYI, these are commercial units and are VERY LOUD. basement or attic installation is a must if you don’t have a network closet. bonus, the ICX6610 allows you to turn on/off POE per port and even set the voltage. this means you can power everything from cameras, stream boxes, wireless APs, to even some POE lights. it’s also a managed switch, which means you can split your ports up on VLANs and other really neat things.
Hi there ! If your i5 gen10 can only handle one 4k transcode, you might want to double check if your hardware acceleration parameters are correctly set ^^
My home server runs on a small Intel N100, a 4k transcode eats up 30% of the CPU at most. I checked, it can transcode 3 streams without any lag, up to 4 transcodings if nothing else is eating the cpu. I’m not using Plex though, I got Jellyfin. I don’t know how Plex handles things.
my i5 doesn’t have a built in graphics core. it’s just a CPU. this means hardware acceleration for video encoding is intensive.
like I said before too, there are other services that are running on the server as well. things like siem, aggregators, containerized desktops, etc.
thanks for the info though! if I ever upgrade the CPU in the future I’ll definitely get one with a GPU for hardware acceleration.
random resource - lots of options. look up *ARR stacks
https://github.com/Rick45/quick-arr-Stack
LLMs
dont bother unless youre shelling out for a big ol gpu etc
Or you can go straight to the source and use the Trash Guides
You can run stuff on a little GPU, IGPU, or even just a CPU. You’ll be limited by RAM/VRAM, and obviously it’ll be slower, but it’ll work.
I have a server that I wanted to do this with, and cpu-only interference was several minutes for pretty simple tasks (all text, couple of sentences). I shelled out the cash for a 5060 Ti 16GB earlier this summer, and even with a pcie 3 x16 slot running at x4, I’m down to 5-30 second responses and rapid output speeds, even with large conversations. But the initial prompt/loading is still a couple of minutes by itself, so I leave the model in vram always.
I can speak to this because one of my old servers was pretty much the machine you’re proposing.
Briefly, in terms of distros, choose what you like (within reason). I’ve always trended towards dietpi, which is extremely light. I actually just stood it up on the Raspberry Pi 4B last night. Headless and by itself it takes a whole whopping 300MB of RAM at idle. Even with all this running,
Synapse
PostgreSQL
Cinny
Arcane
qBittorrent
RDT-Client
Radarr
Sonarr
Prowlarr
Docker
Dropbear
Fail2ban
Tailscale
Syncthing
Argon One
TigerVNC
it sits under a GB
Do your own research, etc.
Pushing 4k streams, it’s not overkill at all, but by the sounds of it, you’re meaning to transcode on the fly? Be aware of AXV-512 support issues etc in Linux vs diff chips. Actually better make sure that your i5 has AXV-512 (not that it’s particularly required for transcoding but it helps) so double check.
Yes, you can run reasonable LLMs on that machine, though without a GPU, you’ll need to investigate exactly which backend runner best supports your CPU.
If you’re looking at a dense model on that rig, consider something in the 4 to 9 billion parameter range. Most comfortably, something like a Qwen3.5-4B, Gemma-4-E4B, MiniCPM or Tiny Ling…or a larger MoE (I like grug 35b-a3b, but Arcee Trinity mini or even GPT-OSS 20B could work) but the more parameters the slower it’ll run. OTOH, just more params =! always better. See -
https://reddit.com/comments/1w47xpd
Gaming wise, obviously you want something that can run Proton, wine, DOSBox, etc. I can’t claim that my pet distro runs all of those OOB because I’ve never tried, but it’s been my general experience that Ubuntu variants seem to do gaming best.
16GB ram is plenty for everything but getting into LLM’s.
If you do that then more is always better… 32GB plus a 6-8GB GPU would be a basic setup for qwen 35b-a3b for exampleStrongly recommend looking into Proxmox hypervisor as your base OS. Then install as many Linux VMs/LXCs as you like, with backups/snapshots all taken care for you. I’m on an Intel 8th gen CPU and have 3-4 VMs running, including Jellyfin, Home Assistant OS, an assortment of random docker images and a dedicated headless Debian VM for various AI projects. The CPU is more than enough, your 12th gen target will breeze through it. Get more RAM than you think you need - it will become the bottleneck for what you can do, not your CPU. I’m running 64GB and don’t regret it!
For video encoding/decoding like Jellyfin/Frigate, make sure you stick with Intel CPU for hardware codec support via QuickSync, otherwise you’ll need to add a GPU.
Yeah, I wasn’t targeting Intel specifically, that’s just what most/all the refurbished SFF PCs I’ve looked at have been running, and I know Intel is good for HEVC transcoding.
Like all projects, you should think about needs, budget, and time. You’ve got a good list of what you want to run which is a good start. Budget is next, particularly considering the price of hardware these days.
FWIW, I run Jellyfin, Navidrome, the entire arr stack, Forgejo, Paperless, mealio, and Immich on a NAS with 24GB RAM and have never needed more. 8GB would have been a bit tight but probably doable. 16 will be plenty for you.
Anything slightly better than a potato will run all of that except the llm’s. To have a useful llm you need to spend circa £1k on a GPU and even then it won’t come close to Claude or chatgpt. Skip the llm’s for now. Just get yourself a solid little server to do everything else you need. Add dedicated hardware specifically for llm’s in future if you really want to. An old used gaming pc will be plenty for your server. I personally use Debian , but there are loads of suitable options available
Thanks. It sounds like the AI stuff will push my budget over what I’m comfortable with and everything else I want will run well under.
You can run an LLM, but it will be limited. Look into Open Vino Model Server.
I’d recommend a minimal install distro (Ubuntu server, Arch, Gentoo, etc…), then add the things you need from there.
Document every step. <— important
@cerebralhawks@lemmy.dbzer0.com this is the very important bit:
Document every step. <— important
Whatever you do will make sense for about a day, you’ll have a rough idea of what you did the following weekend, after that… all gone.
And, whatever you do won’t be right first time, so you will be editing those notes - all part of the journey
Thanks. I took note the first time I read that, but I appreciate the emphasis.
Not llamacpp?
OVMS + a model that was built for it is far more responsive than llamacpp.
