[Mesa-dev] [PATCH 2/2] i965: Guard GetBufferSubData's streaming memcpy load with USE_SSE41

Kenneth Graunke kenneth at whitecape.org
Sat Aug 12 08:40:59 UTC 2017


On Saturday, August 12, 2017 1:25:25 AM PDT Mauro Rossi wrote:
> On Aug 11, 2017 5:14 PM, "Kenneth Graunke" <kenneth at whitecape.org> wrote:
> 
> On Thursday, August 10, 2017 11:50:30 PM PDT Tapani Pälli wrote:
> > I do wonder what the target machine is (I haven't seen one that would
> > not have ARCH_X86_HAVE_SSE4_1 true, both 32bit and 64bit) but falling
> > back to memcpy makes perfect sense without USE_SSE4_1;
> >
> > Reviewed-by: Tapani Pälli <tapani.palli at intel.com>
> 
> I agree - you really want SSE 4.1 support for decent performance on Atoms.
> The MOVNTDQA texture upload path, for example, gave us some really big
> performance gains on some ChromeOS workloads.
> 
> So, while we should patch this, it'd still be worth figuring out if it's
> possible to enable it in Android-x86.
> 
> --Ken
> 
> 
> Hi,
> 
> I checked on nougat-x86 build 32bit and the patches are Ok.
> 
> The users of android-x86 have the option to build with silvermont.mk and
> later
> to have an optimized build, but we currently try to support SSE3 devices.
> GNU/Linux distributions for desktop, like their kernel, may have that goal
> too.
> 
> Thanks
> M.

Well, right...but we do guard these with runtime checks.  So you should
be able to build in optional SSE 4.1 support but still have such a build
work on SSE3-only hardware...

I'm glad to hear it's an option, at least...
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part.
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20170812/1d95a72d/attachment.sig>


More information about the mesa-dev mailing list