fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-20 11:40:10 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

radeonsi: deduplicate min_esverts code in gfx10_ngg_calculate_subgroup_info

Browse source 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13829>
This commit is contained in:
Marek Olšák 2021-11-17 01:08:12 -05:00 • committed by Marge Bot
parent 9d7ac70ffb
commit cdeecadcb6
1 changed files with 5 additions and 14 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

13
src/gallium/drivers/radeonsi/gfx10_shader_ngg.c
View file

@ -2204,7 +2204,8 @@ bool gfx10_ngg_calculate_subgroup_info(struct si_shader *shader)
unsigned gsprim_lds_size = 0; unsigned gsprim_lds_size = 0;
/* All these are per subgroup: */ /* All these are per subgroup: */
const unsigned min_esverts = gs_sel->screen->info.chip_class >= GFX10_3 ? 29 : 24; const unsigned min_esverts =
gs_sel->screen->info.chip_class >= GFX10_3 ? 29 : (24 - 1 + max_verts_per_prim);
bool max_vert_out_per_gs_instance = false; bool max_vert_out_per_gs_instance = false;
unsigned max_gsprims_base = gs_sel->screen->ngg_subgroup_size; /* default prim group size clamp */ unsigned max_gsprims_base = gs_sel->screen->ngg_subgroup_size; /* default prim group size clamp */
unsigned max_esverts_base = gs_sel->screen->ngg_subgroup_size; unsigned max_esverts_base = gs_sel->screen->ngg_subgroup_size;
@ -2290,9 +2291,6 @@ retry_select_mode:
max_esverts = MIN2(max_esverts, max_gsprims * max_verts_per_prim); max_esverts = MIN2(max_esverts, max_gsprims * max_verts_per_prim);
/* Hardware restriction: minimum value of max_esverts */ /* Hardware restriction: minimum value of max_esverts */
if (gs_sel->screen->info.chip_class == GFX10)
max_esverts = MAX2(max_esverts, min_esverts - 1 + max_verts_per_prim);
else
max_esverts = MAX2(max_esverts, min_esverts); max_esverts = MAX2(max_esverts, min_esverts);
max_gsprims = align(max_gsprims, wavesize); max_gsprims = align(max_gsprims, wavesize);
@ -2311,15 +2309,8 @@ retry_select_mode:
} while (orig_max_esverts != max_esverts || orig_max_gsprims != max_gsprims); } while (orig_max_esverts != max_esverts || orig_max_gsprims != max_gsprims);
/* Verify the restriction. */ /* Verify the restriction. */
if (gs_sel->screen->info.chip_class == GFX10)
assert(max_esverts >= min_esverts - 1 + max_verts_per_prim);
else
assert(max_esverts >= min_esverts); assert(max_esverts >= min_esverts);
} else { } else {
/* Hardware restriction: minimum value of max_esverts */
if (gs_sel->screen->info.chip_class == GFX10)
max_esverts = MAX2(max_esverts, min_esverts - 1 + max_verts_per_prim);
else
max_esverts = MAX2(max_esverts, min_esverts); max_esverts = MAX2(max_esverts, min_esverts);
} }