ProxBi

(WIP)

Build a server with multiple GPU-powered virtual machines — perfect for families, classrooms, and homelab enthusiasts. I personally used it for a server for my 3 kids so I do not have to buy a separate PC for each of them and learn something new - in the end it was a successful and interesting project.

---

Table of Contents

1. Overview
2. Features
3. Hardware Requirements
4. Quick Start
5. Setup Instructions
6. Templates & Scripts
7. Demo & YouTube Videos
8. Support & Sponsorship
9. License

---

Overview

ProxBi helps you convert one powerful server into multiple independent desktops for gaming, learning, or productivity. Each user gets their own GPU-powered VM accessible via thin clients or remote desktop.

One Server vs Separate Server per User

Feature / Factor	One Server + Thin Clients	Separate Server per User
Hardware Cost	Lower — shared components (MB, CPU, PSU, case) reduce overall cost	Higher — each user needs a full system
GPU Allocation	Multiple GPUs passed through to different VMs	Each user requires their own GPU
Noise	Minimal in user rooms — server can be placed in a locker, basement, or dedicated room	High — each system produces its own noise
Power Consumption	Lower — single system with shared components	Higher — many separate systems
Maintenance	Centralized updates and monitoring	More effort — updates and troubleshooting for each system
Space Requirements	Compact — one server, small thin clients	Large — full-sized PCs for each user
Flexibility / Scalability	High — can add VMs or GPUs as needed	Limited — adding new users means buying more hardware
Initial Setup Complexity	Higher — requires knowledge of Proxmox and GPU passthrough	Lower — plug-and-play desktop PCs
User Experience	Thin client latency may be slightly higher, but manageable	Native experience, zero virtualization overhead

Tip: Placing the server in a basement or closet reduces noise in living areas while keeping thin clients quiet and efficient. Shared components like motherboard, PSU, and case save money compared to building full PCs for each user.

---

Features

• Automated scripts for Proxmox and GPU passthrough
• Ready-to-use VM templates for Windows 11
• Optimized performance for gaming, coding, and creative apps
• Multi-user management for family, classroom, or homelab setups
• Troubleshooting tips and configuration guides included
• Demo & Videos

---

Hardware Requirements

Tip: Think about future expansions

Server:

• Case with enough space for all components
  • Fans and airflow to keep it cool
  • noise/quietness grade depending on where the server will be placed
• PSU with enough power
• Motherboard with enough and correct ports/pins
• CPU with IOMMU support and enough cores for each user/VM.
• 1+ GPUs for passthrough. One for each user/VM.
• Sufficient RAM (recommend ≥14GB per user/VM).
• Proxmox VE 9.x compatible server
• Enough storage. Preferable SSD or M2.

Clients:

• Thin clients or remote desktop clients for users (mini-PC, laptop, PC, Mac, Mobile etc.)
• Windows 10/11 , Mac, IOS, Android
• 8GB RAM

---

Quick Start

(WIP - proxmox templates or repo)

Setup Instructions

Setup server

1. Enable IOMMU / VT-d in BIOS then reboot

2. Install Proxmox on your server

   • Finally, you should be able to access your Proxmox console

3. Create Windows VMs

   • Use UEFI, Q35 and VirtIO drivers
   • Install the VirtIO ISO for network and disk drivers
   • Install virtual audio driver e.g. https://vb-audio.com/Cable/

4. Passthrough GPUs to VMs. On the proxmox server cli (either in proxmox console>pve node>terminal or ssh into your server) execute:

   • nano /etc/default/grub and edit the line
     • for AMD GRUB_CMDLINE_LINUX_DEFAULT="quiet intel_iommu=on iommu=pt"
     • for Intel GRUB_CMDLINE_LINUX_DEFAULT="quiet amd_iommu=on iommu=pt"
   • update-grub
   • reboot
   • nano /etc/modules and add these lines if not already present
     • vfio
     • vfio_pci
     • vfio_iommu_type1
     • vfio_virqfd
   • lspci -nn | grep -E "VGA|Audio" and note the ids of the GPU VGA and Audio device pairs of each GPU. Something like 01.00.0 and 01:00.1 also 10de:1f08 and 10de:10f9
   • nano /etc/modprobe.d/vfio.conf and add options vfio-pci ids=10de:1f08,10de:10f9 replace with your ids, add all pairs.
   • nano /etc/modprobe.d/blacklist.conf and add:
     • blacklist nouveau
     • blacklist nvidia
     • blacklist nvidiafb
     • blacklist nvidia_drm
   • update-initramfs -u -k all
   • Repeat this for each VM and GPU pair (VMID is the id you gave your VM upon creation) :
     • nano /etc/pve/qemu-server/<VMID>.conf and add the following lines replacing the ids with your GPU id
       • hostpci0: 01:00.0,pcie=1,x-vga=1
       • hostpci1: 01:00.1
   • reboot
   • Login to VM (using proxmox console or parsec) and install the GPU driver from the official provider. You shuld see the GPU in the device manager.

5. Install Parsec and configure as host. For each thin client (mini-pc or laptop) login into VM and then:

   • Download from parsec.app
   • Open and login to parsec. (use separate accounts if you want to separate which VMs each thin client can access)

Setup Thin clients

1. If you have no OS on your client yet : install OS. I used Windows 11.
2. Install Parsec on your thin clients
3. Login to your VM via Parsec

Wake-on-LAN (WoL)

Wake-on-LAN (WoL) is a networking standard that allows a computer to be turned on or woken from a low-power state by a special network message called a "magic packet". To use it, the target computer's motherboard and network adapter must support WoL, and the computer must remain connected to a power source, even when shut down. The feature must also be enabled in the computer's BIOS/UEFI and the network adapter's settings.

1. Enable WoL

   1. from BIOS. Depending on your Motherboard enter BIOS and:
      • enable the option like Power On By PCI-E / PCI
      • and disable ErP Ready (this one cuts power to NIC when server stopped)
   2. in proxmox linux (if not already):
      • ethtool <your-nic> You should see Supports Wake-on: pumbg Wake-on: g if not :
        • edit nano /etc/network/interfaces and add these 2 lines under your interface <your nic>:
          • post-up /usr/sbin/ethtool -s eno1 wol g
          • post-down /usr/sbin/ethtool -s eno1 wol g

2. Use a WoL client

   • Windows
     • GUI
       • From MS Store search for any WoL or MagicPacket client e.g. https://apps.microsoft.com/detail/9wzdncrcw1mx (tested by me on Windows 10/11)
       • https://github.com/basildane/WakeOnLAN/releases/tag/2.12.4 (tested by me on Win 10)
     • PowerShell
       • https://gist.github.com/alimbada/4949168
       • You can make a shortcut in windows
   • Linux
     • GUI
       • https://github.com/muflone/gwakeonlan/ (also from App Center) (tested by me on Ubuntu 24)
       • Ampare Wake on LAN (from App Center) (did not open on my Ubuntu also size too big ~109MB)
     • Terminal
       • wakeonlan (tested by me on Ubuntu 24)
         • sudo apt install wakeonlan
         • wakeonlan <mac address of the server>