I’m thinking I need to plan for the future in a better way then I have been doing. I need a stable distro as a host system that I don’t touch after I have fixed with pass-through settings and vm’s. I was thinking, proxmox would be interesting, but it seems I can’t use my computer and need to access it from another, but if a gpu is on a passthrough it shouldn’t be a problem to access the system at all.
I have a bunch RPi4’s that I could use to connect to the VM machine, but is it going to be possible to game on something that you do a remote connection to? Isn’t the imputlag going to basically kill me at every turn?
Most stable system to act as a VM host? Recommendations?
Huh?
I am using Proxmox exactly like that, proxmox running headless, keyboard, mouse,GPU and sound card passed through an OSX VM, another keyboard/mouse/sound card, gpu passed on to a windows VM …
Your monitor needs to support multiple inputs, and you need a way to start/stop/restart the headless server and its VMs, I use a streamdeck, but a laptop with a web browser or a phone will work as well …
Don’t see the problem either. You can create a VM in Proxmox with passthrough and you can use any self hosting streaming service (from Steam, Nvidia Shadowplay, Parsec etc.). You can also set this VM to autostart on Startup, so its always awayable when you start your PC. Usually you don’t have to interact much with the host OS at all.
The lag is dependent on the stream (1080p vs 4K etc.), your network and connection type (wireless vs. wired) and the clients capability to decode and display the stream on-time.
You’ve not said what computer you have.
Normally people use a separate computer as a Proxmox server. You’d need a reasonably good one, a RYZEN 7 1700 with 32GB RAM is what I’ve got. You need SSDs for your VMs and HDD for your file storage.
If you only have one computer you could do the following. Yes you would need to borrow a laptop or something for the WebGUI whilst you set it up.
First install Proxmox on your new boot drives and set up your new SSDs and HDD for general Proxmox use.
Then using the Proxmox GUI on the laptop create a VM and pass through your USB mouse and keyboard, GPU, PC C: drive to the VM.
Then start the VM. You should see your PC boot up as a VM.
Set the VM to auto boot.
Now you can stop using the laptop and use your VM PC to log into the Proxmox GUI.
Your PC is still a PC but now also a Proxmox hyperviser where you can spin up multiple VMs and LXCs to your hearts content.
There are extensive instructions in the Proxmox documentation for doing what I described above.
I have a bunch of raspberry pi’s so I can probably use that to start and change proxmox. I’m however thinking of buying unRaid, since it seems to be pinning the cpu cores, and I have no Idea if proxmox does that.
I have bots on my computer that I’m thinking might be a bit of a problem to set up. Nadekobot seems to have it’s own docker which means it shouldn’t be much of a problem, but the other one’s need a bunch of applications installed in order to work. I don’t know how much cpu it needs either.
While proxmox/unraid/freenas and other such hyper-visors are interesting, I’m still thinking about having an actual full OS as a host where I have the bots installed, streaming software and such, then have multiple vm’s that I use as my daily drivers, but then again, I don’t want the systems to be slowed down because I have a full host operating system in the background.
It’s the darn bots though. I did have them on a raspberry pi to begin with, but ended up setting them up on my computer again and ditching the raspberry pi, since I didn’t know how long the memory cards would hold up.
Now I have an SSD on one of them, so It could theoretically be used for the bots again.
Is a hypervisor OS better to use as your main computer? Isn’t it loaded into the memory? Doesn’t that mean that I have to, well, have lots more memory?
I would want to use cpu cores specifically with a VM, and they are set as 0-15, so they shouldn’t be too hard to pin to a VM. It’s going to make the system a heck lot more stable also, but that depends on how I set things up. Right?
Stability ‘usually’ has nothing to do with pinning, what pinning will give you is consistent latency and/or more performance when other VMs/tasks are running on the hypervisor.
In the case of audio peripherals,pinning can give more stability in the sense that it may solve crackling/dicsonnects, but that depends on the architecture/audio interface and each case is different imho
Pinning is not hard, but in proxmox you need to configure hooks and make sure you are pinning the right ones. Whic ones depends on the topology of your processor, that may as well not be 0-16, for me, for example, same pair of threads are 0,24,1,25,2,26,3,27,4,28,5,29,6,30, and in my case (EPYC cpu) I try and make sure I am using threads from the same CCD, if possible …
Proxmox itself hardly uses any CPU. Your VMs are capable of using as much CPU as you give them. For practical purposes there is no difference in performance of a VM than bare metal.
Say your RYZEN 7 has 8 cores, that shows up as 16 cores in Proxmox. You can allocate each VM 16 cores if you want, you’re not dividing them up. I would suggest that you always allocate at least 2 cores per VM because operating systems expect 2 or more cores.
Also worth noting for best performance for a VM you can select what type of CPU features to present. It hides things like AVX otherwise.
