Arithmetic throughput
Arithmetic throughput is the rate of arithmetic operations, commonly expressed as floating-point operations per second. Its interpretation depends on the operation and numerical format.
How it works
A fused multiply-add is conventionally counted as two FLOPs. Matrix-operation peaks can differ from scalar-operation peaks, and structured sparsity can introduce a different counting convention. Useful model work can also differ from all instructions a kernel executes.
Work through an example
An illustrative operation containing 2 × 10¹² FLOPs completed in 0.02 seconds delivers 100 TFLOP/s under that work convention.
Where the term can mislead
Compare dense with dense, precision with matching precision, and measured work with a documented numerator. A single peak number cannot describe every kernel.
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.