[Mesa-dev] [PATCH v2 08/13] i965: Handle TCS outputs and TES inputs.

[Kenneth Graunke]

On Saturday, December 12, 2015 10:21:19 PM Jordan Justen wrote:
> On 2015-12-11 13:23:57, Kenneth Graunke wrote:
> > TCS outputs and TES inputs both refer to a common "patch URB entry"
> > shared across all invocations.  First, there are some number of
> > per-patch entries.  Then, there are per-vertex entries accessed via
> > an offset for the variable and a stride times the vertex index.
> > 
> > Because these calculations need to be done in both the vec4 and scalar
> > backends, it's simpler to just compute the offset calculations in NIR.
> > It doesn't necessarily make much sense to use per-vertex intrinsics
> > afterwards, but that at least means we don't lose the per-patch vs.
> > per-vertex information.
> > 
> > Signed-off-by: Kenneth Graunke <kenneth at whitecape.org>
> > ---
> >  src/mesa/drivers/dri/i965/brw_nir.c | 122 +++++++++++++++++++++++++++++++++++-
> >  1 file changed, 120 insertions(+), 2 deletions(-)
> > 
> > diff --git a/src/mesa/drivers/dri/i965/brw_nir.c b/src/mesa/drivers/dri/i965/brw_nir.c
> > index 9cf4944..e46e177 100644
> > --- a/src/mesa/drivers/dri/i965/brw_nir.c
> > +++ b/src/mesa/drivers/dri/i965/brw_nir.c
> > @@ -133,6 +133,69 @@ remap_inputs_with_vue_map(nir_block *block, void *closure)
> >     return true;
> >  }
> >  
> > +struct remap_patch_urb_offsets_state {
> > +   nir_builder b;
> > +   struct brw_vue_map vue_map;
> > +};
> > +
> > +static bool
> > +remap_patch_urb_offsets(nir_block *block, void *closure)
> > +{
> > +   struct remap_patch_urb_offsets_state *state = closure;
> > +
> > +   nir_foreach_instr_safe(block, instr) {
> > +      if (instr->type != nir_instr_type_intrinsic)
> > +         continue;
> > +
> > +      nir_intrinsic_instr *intrin = nir_instr_as_intrinsic(instr);
> > +
> > +      bool is_input =
> > +         intrin->intrinsic == nir_intrinsic_load_input ||
> > +         intrin->intrinsic == nir_intrinsic_load_per_vertex_input;
> > +
> > +      bool is_output =
> > +         intrin->intrinsic == nir_intrinsic_load_output ||
> > +         intrin->intrinsic == nir_intrinsic_load_per_vertex_output ||
> > +         intrin->intrinsic == nir_intrinsic_store_output ||
> > +         intrin->intrinsic == nir_intrinsic_store_per_vertex_output;
> 
> Can you call the functions you added previously? (i965: Separate base
> offset/constant offset combining from remapping.)
> 
> 7 & 8 Reviewed-by: Jordan Justen <jordan.l.justen at intel.com>

Oh, good call - I wrote this patch first, even though it came later in
the series.  That makes a ton of sense.  I've changed it to:

+      gl_shader_stage stage = state->b.shader->stage;
+
+      if ((stage == MESA_SHADER_TESS_CTRL && is_output(intrin)) ||
+          (stage == MESA_SHADER_TESS_EVAL && is_input(intrin))) {
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20151214/f420b263/attachment.sig>