AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Revert "nir/lower_task_shader: don't use base index for shared memory intrinsics"

Browse Source 
This reverts commit e5970fe22a.

Intel backend has implemented the missing functionality.

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Acked-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17618>
This commit is contained in:
Marcin Ślusarz
2022-07-15 13:43:21 +02:00
committed by Marge Bot
parent 66bc9aec65
commit 9f3eb63878
1 changed files with 6 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/compiler/nir/nir_lower_task_shader.c
+6 -4
View File
@@ -55,7 +55,8 @@ lower_nv_task_output(nir_builder *b,
case nir_intrinsic_load_output: {
b->cursor = nir_after_instr(instr);
nir_ssa_def *load =
nir_load_shared(b, 1, 32, nir_imm_int(b, s->task_count_shared_addr));
nir_load_shared(b, 1, 32, nir_imm_int(b, 0),
.base = s->task_count_shared_addr);
nir_ssa_def_rewrite_uses(&intrin->dest.ssa, load);
nir_instr_remove(instr);
return true;
@@ -64,7 +65,8 @@ lower_nv_task_output(nir_builder *b,
case nir_intrinsic_store_output: {
b->cursor = nir_after_instr(instr);
nir_ssa_def *store_val = intrin->src[0].ssa;
nir_store_shared(b, store_val, nir_imm_int(b, s->task_count_shared_addr));
nir_store_shared(b, store_val, nir_imm_int(b, 0),
.base = s->task_count_shared_addr);
nir_instr_remove(instr);
return true;
}
@@ -84,7 +86,7 @@ append_launch_mesh_workgroups_to_nv_task(nir_builder *b,
*/
b->cursor = nir_before_cf_list(&b->impl->body);
nir_ssa_def *zero = nir_imm_int(b, 0);
nir_store_shared(b, zero, nir_imm_int(b, s->task_count_shared_addr));
nir_store_shared(b, zero, zero, .base = s->task_count_shared_addr);
nir_scoped_barrier(b,
.execution_scope = NIR_SCOPE_WORKGROUP,
@@ -104,7 +106,7 @@ append_launch_mesh_workgroups_to_nv_task(nir_builder *b,
.memory_modes = nir_var_mem_shared);
nir_ssa_def *task_count =
nir_load_shared(b, 1, 32, nir_imm_int(b, s->task_count_shared_addr));
nir_load_shared(b, 1, 32, zero, .base = s->task_count_shared_addr);
/* NV_mesh_shader doesn't offer to choose which task_payload variable
* should be passed to mesh shaders, we just pass all.