A workaround described by MakeUseOf lets some older NVIDIA GeForce cards act as multiple virtual GPUs for virtual machines. It could give homelab owners a new use for spare hardware, but the method has a firm generation limit and takes more tinkering than a quick driver install.
NVIDIA Watch analysis
What happened
The article’s author split a GTX 1050 Ti into virtual GPUs by patching NVIDIA’s Linux driver setup and spoofing the card’s identity as a Quadro or Tesla. That allowed the system to expose virtual GPUs for allocation to multiple virtual machines. The MakeUseOf guide describes the process as a hands-on experiment, not an NVIDIA-supported feature.
The demonstrated setup used NVIDIA GRID driver 535.104.06 and a Linux 6.1 kernel. The article says the workaround is limited to cards up to the RTX 2000 series; Ampere and newer cards are not supported by this method. Its GTX 1050 Ti had just 4GB of VRAM, a tight shared pool once divided between virtual machines.
Why it matters
For someone with a compatible older card gathering dust, virtual GPUs could support GPU-accelerated virtual machines without buying a professional card. That is useful for homelabs and experimentation, not a reason to expect a budget gaming GPU to become a modern workstation overnight.
Compatibility is the catch with a capital C: newer consumer cards are out, and the tested setup needed an older kernel, specific drivers and several patches. The card’s memory is shared, too, so splitting it does not make more of it appear.
Our read
This is a plausible weekend project for technically confident owners of older hardware, especially if they already have a suitable card and want to learn about virtualisation. Treat it as an unofficial, hardware-specific workaround, not a supported NVIDIA capability or a drop-in replacement for a professional GPU. If your card is newer than the RTX 2000 generation, this particular route is a dead end.
What to watch
- Whether community tools expand compatibility to newer GPU generations.
- Which older cards offer enough VRAM to make splitting worthwhile.
- Whether future driver and kernel changes make the demonstrated setup harder to maintain.
Discussion spark: Would you try an unofficial workaround on an older GPU, or is supported hardware worth the extra cost?
Sources and evidence
- I unlocked Nvidia’s professional GPU features on my consumer graphics card – MakeUseOf (2 October 2026, 19:45 UTC)
Independent WittyWires tracker for public updates about NVIDIA. Not affiliated with or endorsed by NVIDIA; this is not an official account.