Revert "radv: Disable NGG for GS with suboptimal output vertex count."
It breaks too many things and shouldn't have been merged. The fix isn't
trivial and it will probably not be backported because it's intrusive.
It will be re-applied later when everything will work.
This reverts commit 94706601fa
.
Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15882>
This commit is contained in:

committed by
Marge Bot

parent
e466d73368
commit
2b688942c1
@@ -3206,19 +3206,6 @@ radv_fill_shader_info(struct radv_pipeline *pipeline,
|
||||
stages[MESA_SHADER_TESS_EVAL].info.is_ngg = false;
|
||||
}
|
||||
|
||||
if (stages[MESA_SHADER_GEOMETRY].nir &&
|
||||
stages[MESA_SHADER_GEOMETRY].nir->info.gs.vertices_out >= 9) {
|
||||
/* GS has suboptimal number of output vertices. In this case,
|
||||
* the occupancy of NGG GS is very low, and API GS invocations
|
||||
* can't even occupy a single Wave32 wave.
|
||||
* Therefore the legacy pipeline performs better here.
|
||||
*/
|
||||
if (stages[MESA_SHADER_TESS_EVAL].nir)
|
||||
stages[MESA_SHADER_TESS_EVAL].info.is_ngg = false;
|
||||
else
|
||||
stages[MESA_SHADER_VERTEX].info.is_ngg = false;
|
||||
}
|
||||
|
||||
gl_shader_stage last_xfb_stage = MESA_SHADER_VERTEX;
|
||||
|
||||
for (int i = MESA_SHADER_VERTEX; i <= MESA_SHADER_GEOMETRY; i++) {
|
||||
|
Reference in New Issue
Block a user