Nvidia GPU VRAM used as Linux swap space
The nbd-vram project lets Linux users mount Nvidia GPU VRAM as a swap device, offering a fast‑memory alternative for RAM‑starved systems.
On June 2 2026 a GitHub repository named nbd‑vram appeared, authored by tanelpoder, that implements a network‑block‑device driver allowing Linux to treat a slice of Nvidia GPU VRAM as a swap device [GitHub]. The submission quickly rose to a score of 126 on Hacker News, indicating strong community interest [HN].
What it does
The tool creates a block device backed by GPU memory and registers it with the kernel’s swap subsystem. Because VRAM operates at speeds far above typical SSDs, swapping to it can reduce the latency penalty that conventional disk‑based swap incurs. The project ships a simple command‑line interface and works with any recent Linux kernel that supports the nbd driver.
Why it matters
- Speed advantage – VRAM’s bandwidth and latency are an order of magnitude better than SATA or even NVMe SSDs, so swap I/O can complete much faster.
- Memory relief – Systems with limited RAM can offload inactive pages to the GPU, freeing RAM for active workloads without adding physical memory.
- Linux flexibility – The implementation showcases how the open‑source ecosystem can repurpose existing hardware for unconventional roles, expanding the toolbox for performance‑oriented users.
The project is open‑source, so anyone can build and test it on a compatible Nvidia card. Its adoption will depend on real‑world benchmarks and stability reports from the community.
Subscribe to the broadcast.
Daily digest of the day's most important tech news. No fluff. Engineering signal only.
// delivered via substack · double-opt-in confirmation