libdevice User's Guide :: CUDA Toolkit Documentation

3.140. __nv_fma

Prototype:

double @__nv_fma(double %x, double %y, double %z)

Description:

Compute the value of $x \times y + z$ as a single ternary operation. After computing the value to infinite precision, the value is rounded once.

Returns:

Returns the rounded value of

x \times y + z

as a single operation.

__nv_fma( $\pm \infty$ , $\pm 0$ , z) returns NaN.
__nv_fma( $\pm 0$ , $\pm \infty$ , z) returns NaN.
__nv_fma(x, y, $- \infty$ ) returns NaN if $x \times y$ is an exact $+ \infty$ .
__nv_fma(x, y, $+ \infty$ ) returns NaN if $x \times y$ is an exact $- \infty$ .

Note:

For accuracy information see the CUDA C++ Programming Guide, Mathematical Functions Appendix, Double-Precision Floating-Point Functions section.

Library Availability:

Compute 2.0: Yes

Compute 3.0: Yes

Compute 3.5: Yes