someone bolted a $266 Nvidia server GPU to their gaming PC. it's loud as a lawnmower — and it works

A used Tesla V100 and a $50 adapter got one builder to 32GB of VRAM for $266 total — here's the honest math on whether you should copy them. Aliteq Ravi Malhotra · Hardware Editor

A Tesla V100 SXM2 16GB plus a bare SXM2-to-PCIe adapter added 16GB of VRAM to an RTX 4080 gaming PC for $266 total, for 32GB combined.

The rig ran Qwen3.6-27B (Q5_K_M) at roughly 32 tokens per second, with prompt processing around 133–160 tokens per second.

Stock cooling hit 82 decibels under load — tunable down with fan-curve work, but the card still needs real airflow.

The V100 has no display output of its own and no driver support past Nvidia's legacy 535 branch, so it can't run alone.

Power draw peaked around 150W — less than half what a single RTX 3090 pulls under a full load.

A clever, cheap way to add raw VRAM to a system that already has a working display GPU — not a card to build a first local-AI rig around. The math only wins if you value memory over convenience,…

Read the full story

Aliteq someone bolted a $266 Nvidia server GPU to their gaming PC. it's loud as a lawnmower — and it works Read the full story on Aliteq