[Development] QTBUG-30440: restricting the SIMD files

Oswald Buddenhagen oswald.buddenhagen at digia.com
Fri Aug 16 13:44:26 CEST 2013


On Thu, Aug 15, 2013 at 11:52:34AM -0700, Thiago Macieira wrote:
> On quinta-feira, 15 de agosto de 2013 20:20:04, Oswald Buddenhagen wrote:
> > intel atoms are based on P5 (are they still?),
> 
> Atoms are actually based on the Pentium-M, which is an offshot of the
> Pentium III.
> 
hmm, seems like the early reports i read about it were misleading.
http://en.wikipedia.org/wiki/Bonnell_%28microarchitecture%29 suggests
that the architecture is something like a "castrated P6", which is
consistent with what you said. you may want to add some more info to
that page.

anyway, the precise architecture and its "genealogy" don't really matter
for the implemented instruction set, which is what matters to us.

> In fact, all Atoms have support for SSSE3 (same level as Core2),
> which AMD CPUs didn't have until 2011.
> 
> > P6+sse2 (over a decade old) is *probably* a good default for desktop
> > builds. ^^
> 
> Agreed,
>
good

> but one of our most interesting optimisations in the draw helpers is
> to use the PALIGNR instruction, which came with SSSE3. We still need
> runtime detection then.
> 
it can't be too easy, can it? ^^

> > it would probably make most sense to have configure set some
> > target-specific optimization defaults (for the embedded devices it would
> > be possible to put them directly into the qmake specs).
> 
> Just set the -march= that you want to set and it should be fine.
> 
yeah.
which reminds me that the CFLAGS & co. handling in configure and qmake
suck. ehm.



More information about the Development mailing list