[Mesa-dev] [PATCH 7/8] i965/msaa: Add CMS support to blorp.

[Paul Berry]

On 11 July 2012 11:16, Paul Berry <stereotype441 at gmail.com> wrote:

> On 11 July 2012 11:01, Chad Versace <chad.versace at linux.intel.com> wrote:
>
>> On 07/06/2012 03:29 PM, Paul Berry wrote:
>> > This patch updates the blorp engine to properly handle the case where
>> > the surface being textured from uses Gen7's CMS MSAA layout.  The
>> > following changes were necessary:
>> >
>> > - Before reading color values from the surface, we need to read from
>> >   the MCS buffer using the ld_mcs sampler message.  This is done by
>> >   the mcs_fetch() function, and the result is stored in the mcs_data
>> >   register.  This only needs to be done once per pixel, since the MCS
>> >   value is shared between all samples belonging to a pixel.
>> >
>> > - When reading color values from the surface, we need to use the
>> >   ld2dms sampler message instead of the ld2dss message, and we need to
>> >   provide the value read from the MCS buffer as an argument.
>> > ---
>> >  src/mesa/drivers/dri/i965/brw_blorp_blit.cpp |   45
>> ++++++++++++++++++++++++-
>> >  1 files changed, 43 insertions(+), 2 deletions(-)
>>
>> I see an easy optimization that's suitable for a small follow-up patch.
>> But this
>> code works and passes tests, so let's commit it as-is first.
>>
>
> Cool.  I'm curious to hear what your optimization idea is.
>

Oh!  The optimization you are referring to is the one you mention later in
the email!  I get it now.


>
>
>>
>> This patch is
>> Reviewed-by: Chad Versace <chad.versace at linux.intel.com>
>>
>> > @@ -1126,8 +1148,13 @@ brw_blorp_blit_program::texel_fetch(struct
>> brw_reg dst)
>> >        break;
>> >     case 7:
>> >        if (key->tex_samples > 0) {
>> > -         texture_lookup(dst, GEN7_SAMPLER_MESSAGE_SAMPLE_LD2DSS,
>> > -                        gen7_ld2dss_args,
>> ARRAY_SIZE(gen7_ld2dss_args));
>> > +         if (key->tex_layout == INTEL_MSAA_LAYOUT_CMS) {
>> > +            texture_lookup(dst, GEN7_SAMPLER_MESSAGE_SAMPLE_LD2DMS,
>> > +                           gen7_ld2dms_args,
>> ARRAY_SIZE(gen7_ld2dms_args));
>> > +         } else {
>> > +            texture_lookup(dst, GEN7_SAMPLER_MESSAGE_SAMPLE_LD2DSS,
>> > +                           gen7_ld2dss_args,
>> ARRAY_SIZE(gen7_ld2dss_args));
>> > +         }
>>
>> The hw docs suggest here "a simple optimization with probable large
>> return in
>> performance". If mcs_data == 0, which should be the common case for most
>> fragments, then only sample array slice 0 needs to be accessed. That is,
>> when
>> tex_layout is CMS then we should emit code like this here:
>>
>>   if (mcs_data == 0) {
>>     ld2dss(S, X, Y);
>>   } else {
>>     ld2dms(S, mcs_data, X, Y);
>>   }
>>
>
> I actually believe this will be slightly faster:
>
> ld2dms(0, mcs_data, X, Y);
> if (mcs_data != 0) {
>   for (S = 1; S < num_samples; ++S)
>     ld2dms(S, mcs_data, X, Y);
>     average samples together;
>   }
> }
>
> Because it will avoid issuing an extra sample instruction in the case
> where some pixels have a zero value for mcs_data and others have a nonzero
> value for mcs_data.  I've got that queued up for a patch series after this
> one is reviewed :)
>
>
>>
>> ----
>> Chad Versace
>> chad.versace at linux.intel.com
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20120711/a56d9e40/attachment.html>