[Mesa-dev] [PATCH 2/6] i965/fs: Track the maximum surface index used in brw_wm_prog_data.

[Kenneth Graunke]

On Friday, August 16, 2013 06:11:25 AM Paul Berry wrote:
> On 14 August 2013 21:07, Kenneth Graunke <kenneth at whitecape.org> wrote:
> > This allows us to determine how small we can make the binding table.
> > 
> > Since it depends entirely on the shader program, we can just compute
> > it once at compile time, rather than at binding table emit time (which
> > happens during drawing).
> > 
> > Signed-off-by: Kenneth Graunke <kenneth at whitecape.org>
> > ---
> > 
> >  src/mesa/drivers/dri/i965/brw_context.h   |  2 ++
> >  src/mesa/drivers/dri/i965/brw_fs.h        |  2 ++
> >  src/mesa/drivers/dri/i965/brw_fs_emit.cpp | 23 +++++++++++++++++++++++
> >  3 files changed, 27 insertions(+)
> > 
> > diff --git a/src/mesa/drivers/dri/i965/brw_context.h
> > b/src/mesa/drivers/dri/i965/brw_context.h
> > index ff0a65c..380fe08 100644
> > --- a/src/mesa/drivers/dri/i965/brw_context.h
> > +++ b/src/mesa/drivers/dri/i965/brw_context.h
> > @@ -305,6 +305,8 @@ struct brw_wm_prog_data {
> > 
> >     GLuint reg_blocks_16;
> >     GLuint total_scratch;
> > 
> > +   unsigned max_surface_index;
> > +
> 
> I'm bothered by the off-by-one inconsistency of using max_surface_index
> here, but using binding_table_size over in brw_vec4_prog_data (see patch
> 5).  Could we change this to binding_table_size, and update
> fs_generator::mark_surface_used() to do:
> 
> prog_data->binding_table_size = MAX2(prog_data->binding_table_size,
> surf_index + 1);
> 
> Then it would be consistent with vec4_generator::mark_surface_used().
> 
> With that changed, this patch is:
> 
> Reviewed-by: Paul Berry <stereotype441 at gmail.com>

Sure.  That would be better.

I wrote this code first, and didn't think about the case where there were 0 
surfaces.  For the FS, it doesn't matter much since you always have at least 
one render target at surface index 0.  For the VS, surface index 0 is the pull 
constant buffer, which is very optional.

I'll make them consistent.

--Ken