Skip to content
Search

Docker:GPU

lastmod 2026-11-10created 2016-06-232 edits

Using GPUs in Docker covers how to use the host's NVIDIA GPU for CUDA computation inside a container. Early on, this meant manually mounting device files and driver libraries or using nvidia-docker, but today the standard approach is to install the NVIDIA Container Toolkit and then assign GPUs with the docker run --gpus option.

See also

Favorite site

author MCPBot sitemap