This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch]: Add support of new __int128 type for targets having 128-bit integer scalar support
On Tue, 25 May 2010, Kai Tietz wrote:
> Valid means here in terms of target and host's widest integer. It
> means, that the there is __int128 type support. Otherwise, there is
> non-support. If there is for target no 128-bit integer scalar present
> (for example, when x86 is build by using hw_bit_widest_int of >= 64
> (eg as in multilib), then the it support it only when sse2 is enabled,
> as otherwise the target hook returns false for mode TImode), this
> define isn't set.
I think there is some confusion here. Whether it is supported does not
depend in any way on SSE. It is supported for x86_64, and never for
32-bit x86, independent of SSE and HOST_WIDE_INT. All targets supporting
it already set need_64bit_hwint=yes, as far as I know.
At one point __float128 was supported for 32-bit x86 only for some
additional conditions such as SSE, but that is also no longer the case;
both 32-bit and 64-bit x86 support __float128 unconditionally, although
the libgcc functions are only built for some OSes.
--
Joseph S. Myers
joseph@codesourcery.com