CUDA Runtime API
The CUDA Runtime API supplies common application operations such as allocation, copies, stream management and kernel launch support. Its shared-library form on Linux is libcudart.so.
How it works
The runtime handles several details that an application using the Driver API would manage explicitly, including use of a device’s primary context. Runtime libraries can be distributed with an application, but they still rely on a compatible driver.
Work through an example
A framework wheel may carry its own runtime version while another application on the same host uses a different one. Inspect the process’s loaded libraries when diagnosing an actual mismatch.
Where the term can mislead
libcudart and libcuda are different libraries. Installing a newer toolkit does not automatically upgrade the host’s kernel driver.
Connect the idea
Take it into practice
Continue with the existing concept, apply it to an interview problem, or follow a course lesson. Access requirements are shown before you open a page.
New to GPUs? Start with the free execution-model concept.
Sources and scope
Examples here are teaching calculations, not measured GPU performance. Consult the linked documentation for the exact device, instruction and tool version. Source review: 25 October 2026.