[RFC] Fix PR19401: always completely peel loops

Andi Kleen ak@muc.de
Fri Jan 14 03:14:00 GMT 2005


Richard Guenther <rguenth@tat.physik.uni-tuebingen.de> writes:
>
> Well, it's motivated two-ways - for one, we currently generate very
> much worse code for std::pow(x, 2) than for std::pow(x, 2.0) if you
> do not specify -funroll-loops -- for another, in gcc 3.4 I could get

Sounds like another case crying for a unroll loops pragma.
Then libstdc++ could unroll the loops that needs it without
impacting other code.

-Andi



More information about the Gcc-patches mailing list