Memory-bound
A memory-bound workload makes progress at a rate limited by its memory accesses. The cause can be transfer bandwidth, dependency latency or another constrained part of the memory path.
How it works
Bandwidth saturation needs enough independent requests in flight. A dependent chain of loads may achieve little bandwidth because the next address is unavailable until the previous load finishes. These cases require different changes.
Work through an example
For an illustrative 8 GB transfer at 2 TB/s, the bandwidth-only time floor is 4 ms. A much longer observed duration calls for investigating traffic, concurrency, cache behavior and dependencies.
Where the term can mislead
Low DRAM throughput does not rule out a memory-related delay. Identify both the memory level and the limiting mechanism.
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.