Mercurial > octave-nkf

changeset 8899:4ab9654eceac

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
update comment
author Jaroslav Hajek <highegg@gmail.com>
date Mon, 02 Mar 2009 12:12:29 +0100
parents ca032de5fbf2
children 63ad1133d0ed
files liboctave/mx-inlines.cc
diffstat 1 files changed, 3 insertions(+), 1 deletions(-) [+]
line wrap: on
line diff
--- a/liboctave/mx-inlines.cc	Mon Mar 02 10:22:17 2009 +0100
+++ b/liboctave/mx-inlines.cc	Mon Mar 02 12:12:29 2009 +0100
@@ -286,7 +286,9 @@
 OP_DUP_FCN (imag, mx_inline_imag_dup, float,  FloatComplex)
 OP_DUP_FCN (conj, mx_inline_conj_dup, FloatComplex, FloatComplex)
 
-// NOTE: std::norm is NOT equivalent
+// FIXME: Due to a performance defect in g++ (<= 4.3), std::norm is slow unless
+// ffast-math is on (not by default even with -O3). The following helper function
+// gives the expected straightforward implementation of std::norm.
 template <class T>
 inline T cabsq (const std::complex<T>& c) 
 { return c.real () * c.real () + c.imag () * c.imag (); }