https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83661
Christophe Monat <christophe.monat at st dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |christophe.monat at st dot com
--- Comment #2 from Christophe Monat <christophe.monat at st dot com> ---
Hi Pratamesh,
Just curious to understand how this can be beneficial performance-wise since
most (all ?) implementations of sincos fail to reap real benefit from the
unique call and finally compute separately sin and cos, after some extra
control-flow.
The non-optimized version costs : add (to compute *2.0), sin, sin, add
The optimized version costs (I choose to account for sincos as being sin, then
cos IRL) : sin, cos, mul, mul, add
It looks to me that the situation is not favorable for the optimized version -
at least one more operation.
What do you think ?
