aco: Calculate workgroup size of legacy GS.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4388>

diff --git a/src/amd/compiler/aco_instruction_selection_setup.cpp b/src/amd/compiler/aco_instruction_selection_setup.cpp
index c09d1459846b0baf357d1fefef212a05471df0bb..bf4b34c6e3e9dd5ac05d4e9b3efd68a34d42c1a9 100644 (file)
--- a/src/amd/compiler/aco_instruction_selection_setup.cpp
+++ b/src/amd/compiler/aco_instruction_selection_setup.cpp
@@ -1293,7 +1293,11 @@ setup_isel_context(Program* program,
       program->workgroup_size = program->wave_size;
    } else if (program->stage & hw_gs) {
       /* If on-chip GS (LDS rings) are enabled on GFX9 or later, merged GS operates in workgroups */
-      program->workgroup_size = UINT_MAX; /* TODO: set by VGT_GS_ONCHIP_CNTL, which is not plumbed to ACO */
+      assert(program->chip_class >= GFX9);
+      uint32_t es_verts_per_subgrp = G_028A44_ES_VERTS_PER_SUBGRP(program->info->gs_ring_info.vgt_gs_onchip_cntl);
+      uint32_t gs_instr_prims_in_subgrp = G_028A44_GS_INST_PRIMS_IN_SUBGRP(program->info->gs_ring_info.vgt_gs_onchip_cntl);
+      uint32_t workgroup_size = MAX2(es_verts_per_subgrp, gs_instr_prims_in_subgrp);
+      program->workgroup_size = MAX2(MIN2(workgroup_size, 256), 1);
    } else if (program->stage == vertex_ls) {
       /* Unmerged LS operates in workgroups */
       program->workgroup_size = UINT_MAX; /* TODO: probably tcs_num_patches * tcs_vertices_in, but those are not plumbed to ACO for LS */