This revision was automatically updated to reflect the committed changes.
Closed by commit rL271218: [X86][SSE] _mm_store1_ps/_mm_store1_pd should
require an aligned pointer (authored by RKSimon).
Changed prior to commit:
http://reviews.llvm.org/D20617?vs=58397=58979#toc
Repository:
rL LLVM
RKSimon added inline comments.
Comment at: lib/Headers/emmintrin.h:598
@@ -594,3 +597,3 @@
static __inline__ void __DEFAULT_FN_ATTRS
-_mm_store_pd(double *__dp, __m128d __a)
+_mm_store_pd1(double *__dp, __m128d __a)
{
majnemer wrote:
> You could use
majnemer added a subscriber: majnemer.
Comment at: lib/Headers/emmintrin.h:598
@@ -594,3 +597,3 @@
static __inline__ void __DEFAULT_FN_ATTRS
-_mm_store_pd(double *__dp, __m128d __a)
+_mm_store_pd1(double *__dp, __m128d __a)
{
You could use
craig.topper accepted this revision.
craig.topper added a comment.
This revision is now accepted and ready to land.
Given that its documented as being aligned. I'm ok with it. LGTM
Repository:
rL LLVM
http://reviews.llvm.org/D20617
___
RKSimon added a comment.
In http://reviews.llvm.org/D20617#439200, @craig.topper wrote:
> Can you double check gcc's xmmintrin.h again. I'm pretty sure _mm_store1_ps
> is calling _mm_storeu_ps.
Yes you're right - for gcc _mm_store1_pd is aligned (and there is a comment
saying it must be),
Can you double check gcc's xmmintrin.h again. I'm pretty sure _mm_store1_ps
is calling _mm_storeu_ps.
On Wed, May 25, 2016 at 3:31 AM, Simon Pilgrim
wrote:
> RKSimon created this revision.
> RKSimon added reviewers: craig.topper, spatel, andreadb.
> RKSimon added a
RKSimon created this revision.
RKSimon added reviewers: craig.topper, spatel, andreadb.
RKSimon added a subscriber: cfe-commits.
RKSimon set the repository for this revision to rL LLVM.
According to the gcc headers, intel intrinsics docs and msdn codegen the
_mm_store1_ps/_mm_store1_pd (and