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

3.267. __nv_scalbn

Prototype:

double @__nv_scalbn(double %x, i32 %y) 

Description:

Scale x by 2 n by efficient manipulation of the floating-point exponent.

Returns:

Returns x * 2 n .
  • __nv_scalbn( ± 0 , n) returns ± 0 .

  • __nv_scalbn(x, 0) returns x.

  • __nv_scalbn( ± , n) returns ± .

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