AI engineering glossary · Computing foundations
What is VRAM?
VRAM (also called Video RAM, GPU memory): The dedicated high-speed memory attached to a GPU, holding the data the GPU is working on, such as model weights, activations and images. If a model does not fit in VRAM, it cannot run on that GPU without tricks such as offloading or compression.
VRAM: an example
A model with 7 billion weights stored at 2 bytes each needs about 14 GB of VRAM just for the weights.
Related terms
More Computing foundations terms
- Wide area network
- x86
- Application programming interface
- ARM architecture
- ASCII
- Asynchronous programming
- Bandwidth
- Base64
Where you learn it
VRAM is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.