This is the mail archive of the
mailing list for the GCC project.
Re: Patch to fix _mm_sqrt_sd intrinsic
On Wed, 2005-08-24 at 20:13 -0700, Richard Henderson wrote:
> On Wed, Aug 24, 2005 at 05:45:20PM -0700, Fariborz Jahanian wrote:
> > The Intel reference manuals say it shall be this:
> > __m128d _mm_sqrt_sd( __m128d );
> Not the one I'm looking at, it doesn't.
> __m128d __mm_sqrt_sd (__m128d a, __m128d)
> Computes the square root of the lower DP FP value of b.
> The upper DP FP value is passed through from a.
> r0 := sqrt(b0)
> r1 := a1
same here for 9.0 (the latest):
__m128d _mm_sqrt_sd(__m128d a, __m128d b)
Computes the square root of the lower DP FP value of b. The upper DP FP
value is passed through from a.
r0 := sqrt(b0)
r1 := a1
(file:///opt/intel/cc/9.0/doc/main_cls/mergedProjects/cref_cls/common/comm1037.htm, for those who installed it in the default place)
> Inconveniently, Intel has put no version number on this document that
> I can find. I *think* it's from the icc 8.0 release, but I'm not sure.
> I can tell you that I downloaded it 2005-01-09. If someone has access
> to all icc releases, sha1 is
> 659d543b59c03d8d0e025f50006e72342f708986 c_ug_lnx.pdf
> That should match some release and we can put an upper bound on the
> latest icc release it could have come from.