[Mesa-dev] [v2 18/19] i965: Add helper for lossless compression support

[Pohjolainen, Topi]

On Mon, Feb 15, 2016 at 01:27:19PM -0800, Ben Widawsky wrote:
> On Thu, Feb 11, 2016 at 08:34:11PM +0200, Topi Pohjolainen wrote:
> > v2: Use explicitly against base type of GL_FLOAT instead of
> >     using _mesa_is_format_integer_color(). Otherwise we miss
> >     GL_UNSIGNED_NORMALIZED.
> > 
> > Signed-off-by: Topi Pohjolainen <topi.pohjolainen at intel.com>
> > ---
> >  src/mesa/drivers/dri/i965/intel_mipmap_tree.c | 22 ++++++++++++++++++++++
> >  src/mesa/drivers/dri/i965/intel_mipmap_tree.h |  3 +++
> >  2 files changed, 25 insertions(+)
> > 
> > diff --git a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> > index 6c233d8..e9fbeeb 100644
> > --- a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> > +++ b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> > @@ -294,6 +294,28 @@ intel_miptree_is_lossless_compressed(const struct brw_context *brw,
> >     return mt->num_samples <= 1;
> >  }
> >  
> > +bool
> > +intel_miptree_supports_lossless_compressed(mesa_format format)
> > +{
> > +   /* For now compression is only enabled for integer formats even though
> > +    * there exist supported floating point formats also. This is a heuristic
> > +    * decision based on current public benchmarks. In none of the cases these
> > +    * formats provided any improvement but a few cases were seen to regress.
> > +    * Hence these are left to to be enabled in the future when they are known
> > +    * to improve things.
> > +    */
> > +   if (_mesa_get_format_datatype(format) == GL_FLOAT)
> > +      return false;
> > +
> > +   /* In principle, fast clear mechanism and lossless compression go hand in
> > +    * hand. However, fast clear can be also used to clear srgb surfaces by
> > +    * using equivalent linear format. This trick, however, can't be extended
> > +    * to be used with lossless compression and therefore a check is needed to
> > +    * see if the format really is linear.
> > +    */
> > +   return _mesa_get_srgb_format_linear(format) == format;
> > +}
> > +
> 
> Hmm. Doesn't this need to use the ccs_e field in surface formats, or did I miss
> something?

It does but I re-used intel_miptree_supports_non_msrt_fast_clear() to check
if a format is supported. There you can see the trick Neil introduced:

if (brw->gen >= 9) {
   mesa_format linear_format = _mesa_get_srgb_format_linear(mt->format);
   const uint32_t brw_format = brw_format_for_mesa_format(linear_format);
   return brw_losslessly_compressible_format(brw, brw_format);
} else ...

None of the srgb formats are supported even for fast clear, but Neil
found out that we can use fast clear in certain cases even for them.

I need to undo the conversion from srgb to linear to know for sure the format
is supported for compression. By the time I wrote this there wasn't direct
utility for checking if format srgb and therefore I chose to write it that
way.