This document includes math equations (highlighted in red) which are best viewed with Firefox version 4.0 or higher, or another MathML-aware browser. There is also a PDF version of this document.
libdevice User's Guide (PDF) - CUDA Toolkit v5.5 (older) - Last updated July 19, 2013 - Send Feedback

3.266. __nv_saturatef

Prototype:

float @__nv_saturatef(float %x) 

Description:

Clamp the input argument x to be within the interval [+0.0, 1.0].

Returns:

  • __nv_saturatef(x) returns 0 if x < 0.

  • __nv_saturatef(x) returns 1 if x > 1.

  • __nv_saturatef(x) returns x if 0 x 1 .

  • __nv_saturatef(NaN) returns 0.

Library Availability:

Compute 2.0: Yes

Compute 3.0: Yes

Compute 3.5: Yes


libdevice User's Guide (PDF) - CUDA Toolkit v5.5 (older) - Last updated July 19, 2013 - Send Feedback