Special function unit
A special function unit executes particular mathematical operations, such as supported approximations used in reciprocal, exponential or trigonometric calculations.
How it works
The compiler can lower a source expression into several instructions. Some may use a special-function pipeline while range reduction, scaling or accuracy corrections use other arithmetic units. The generated instruction sequence and precision mode determine the actual work.
Work through an example
A softmax kernel performs an exponential as well as subtraction, reduction and division. Even when its matrix operations are fast elsewhere, these scalar operations can limit that kernel.
Where the term can mislead
Do not assign every math-library call a single SFU instruction or assume approximate intrinsics preserve the original numerical contract.
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.