I have thought about that, but Proxmox already has built-in a lot of things for virtualization and managing VMs and has less bloat because it has only one purpose.
I have thought about that, but Proxmox already has built-in a lot of things for virtualization and managing VMs and has less bloat because it has only one purpose.
If i mostly use CTs/LXCs the impact should be minimal (in theory), maybe even better if i dont have everything powerd up.
That’s interesting, haven’t considered that. Although I would want to run most things in CTs/LXCs and not full VMs for performance reasons. And Proxmox has more DIY feel which i kind of like. If I fail with Proxmox, might give QubesOS a try.
I would do everything in VMs, mostly Linux and probably one Windows. Proxmox would be only for managing VMs. I want everything in VMs because it’s more flexible for partitioning storage and i can have both Linux and Windows runing at the same time (which can’t be done with dualboot). I am student of computer science so i use it for programming, both for college and side projects. Sometimes there are a lot of programs i have so OS kind of gets bloated, not so much from performance standpoint but just mental overhead of having 10, 20, 30 programs and having to keep in mind what program needs what dependencies, env variables, etc… so i want to kind of group them to VMs and CTs.
its wierd that it is compared by CO2/mass, it would make more sense per callories or some important nutrients.
You definitely shouldn’t overprovision cores (logical cores, 16 in your case), especially for gamming VM. Because you are not going to get more performance, you are only going to increase overhead of context switching. One way to “pin” VM to half of cores is to give your VM 8 vCPU and increase vCPU units value (default is 1000). vCPU units determines VM priority in case of more work than CPU can handle.
https://forum.proxmox.com/threads/cpu-units.1047/
Basicaly if you have 2 VMs with default value of 1000 and CPU can’t handle everything, proxmox will give both VMs equal cpu time (50/50).
But if for example you give first VM 3000 vCPU units, then under heavy load CPU time will be divided 75/25.
Time on CPU will be proportional to given vCPU units (under load).
So if you want to “pin” VM to 8 logical cores, then give VM 8 vCPUs and increase vCPU units (depending how big of priority you want).
True, but if internet connection for some reason crashes in some part of the world you still can’t use crypto there, so cash is still not useless (and probably shouldn’t ever be).