Memory bandwidth: what it means.
Memory bandwidth describes how much data can arrive or move in a given time. Intel distinguishes that transfer rate from latency, the delay between requesting information and receiving it. The two can affect an application differently. (Intel: memory performance)
A rate in GB/s is not a capacity in GB. A high peak transfer specification also does not state that every access pattern sustains that rate. The application may wait for dependencies or use only part of the transferred data. (Intel: memory performance)
Put the definition to work.
Keep storage capacity and transfer rate separate: GB describes an amount of data, while GB/s describes a rate. A peak interface specification is not a promise of sustained application performance. (Micron: high-bandwidth memory)
Questions to ask next.
Use these questions as a reading checklist, not as a device-selection recommendation. Exact manufacturer documentation takes priority over a general definition.
- Units and peak versus measured rate
- Access pattern and request delays
- Per-interface versus aggregate figures
For the broader explanation, explore memory and bandwidth. This page defines the term; the companion guide connects it to the rest of the system.
Common Memory bandwidth questions.
Can high bandwidth eliminate all memory waiting?
No. Latency and the application’s access behavior still matter. Identify whether the workload is constrained by transfer volume, request delays or another part of execution. (Intel: memory performance)