[PATCH] D20358: [Clang][AVX512][Intrinsics]Convert AVX non-temporal store builtins to LLVM-native IR.

[Simon Pilgrim via cfe-commits]

RKSimon added a comment.

In http://reviews.llvm.org/D20358#446241, @RKSimon wrote:

> In http://reviews.llvm.org/D20358#446220, @ab wrote:
>
> > In http://reviews.llvm.org/D20358#446218, @ab wrote:
> >
> > > In http://reviews.llvm.org/D20358#446210, @RKSimon wrote:
> > >
> > > > Is there any reason why we can't just get rid of all the SSE movnt builtins and use __builtin_nontemporal_store instead (http://reviews.llvm.org/D12313)?
> > >
> > >
> > > I wanted to suggest that too, but I think you'd have problems with the (natural?) alignment requirement of __builtin_nontemporal_store (whereas IIRC, movnti & friends accept unaligned pointers).
> >
> >
> > But now that I look at this again, I suppose we could have some __attribute__((aligned(1))), or something like r271214.
>
>
> True, luckily that only affects _mm_stream_si32 and _mm_stream_si64 - the 'real' vector movnt stores all require type alignment. The _mm_stream_load_* (movntdqa) loads cases should be trivial as well.


I've created http://reviews.llvm.org/D21272 that covers the conversion of SSE/SSE2/AVX/AVX512 non-temporal aligned vector stores to use __builtin_nontemporal_store in headers


http://reviews.llvm.org/D20358