[Development] Enabling SSE2 by default on x86 builds (32-bit)

[Thiago Macieira]

On quarta-feira, 11 de dezembro de 2013 01:01:30, Allan Sandfeld Jensen wrote:
> On Tuesday 10 December 2013, Thiago Macieira wrote:
> > [*] note that it's technically SSE support; SSE2 added the integer
> > operations to SSE, which isn't what QtQuick needs.
> 
> SSE2 also added support for double, SSE(1) only had 32bit float support.
> Which is why it is only with SSE2 that you can fully replace x87 with SSE
> in C/C++.

Right, I had never noticed that. I'm mixing with AVX2.

MOVAPS, MOVUPS, MULPS, MULSS were added with SSE.

MOVAPD, MOVUPD, MULPD, MULSD were added in SSE2.

Fortunately, both sets were extended to 256-bit in AVX. Only the 256-bit 
integer instructions are left for AVX2.

AVX512F extends both sets to 512-bit.
-- 
Thiago Macieira - thiago.macieira (AT) intel.com
  Software Architect - Intel Open Source Technology Center
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 190 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.qt-project.org/pipermail/development/attachments/20131210/a1fc06c2/attachment.sig>