# Model bandwidth utilization (MBU)

> Model bandwidth utilization (MBU): The share of peak memory bandwidth inference uses: parameters × bytes per parameter × tokens per second, over the…

Canonical: https://llmdojo.dev/glossary/mbu

Also called MBU.

The share of peak memory bandwidth inference uses: parameters × bytes per parameter × tokens per second, over the peak, for one sequence.

Learn it in the lesson: [Measuring decode and prefill on a roofline](https://llmdojo.dev/learn/inference-optimization/where-the-time-goes/the-roofline) (Inference and Model Optimization · Where the Time Goes: Memory, Compute and Batching).

## Related
- [Compute-bound](https://llmdojo.dev/glossary/compute-bound)
- [Memory bandwidth-bound](https://llmdojo.dev/glossary/bandwidth-bound)
- [Arithmetic intensity](https://llmdojo.dev/glossary/arithmetic-intensity)
- [Roofline](https://llmdojo.dev/glossary/roofline)
- [Continuous batching](https://llmdojo.dev/glossary/continuous-batching)
- [Glossary of every term](https://llmdojo.dev/glossary)