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

mesa: Retire the OptimizeForAOS code.

Browse Source 
It's been unused for years.  Back on the brw driver, it was used for a ~1%
win on a VS-heavy workload, because you could hide instruction latency
using a DP4 sequence instead of MADs for doing the MVP transform.  Given
that r300 and i915 don't seem to want it, and nobody has successfully
ported it for crocus (see
https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14277), just
garbage collect the code at this point.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35617>
This commit is contained in:
Emma Anholt
2025-06-18 14:17:07 -07:00
committed by Marge Bot
parent f7aa6ba906
commit 6808ccf23a
10 changed files with 18 additions and 189 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/compiler/glsl/glsl_parser_extras.cpp
-3
View File
@@ -2507,9 +2507,6 @@ do_common_optimization(exec_list *ir, bool linked,
OPT(do_if_simplification, ir);
OPT(opt_flatten_nested_if_blocks, ir);
if (options->OptimizeForAOS && !linked)
OPT(opt_flip_matrices, ir);
OPT(do_dead_code_unlinked, ir);
OPT(do_tree_grafting, ir);
OPT(do_minmax_prune, ir);
src/compiler/glsl/ir_optimization.h
-1
View File
@@ -44,7 +44,6 @@ bool do_algebraic(exec_list *instructions, bool native_integers,
bool do_dead_code(exec_list *instructions);
bool do_dead_code_local(exec_list *instructions);
bool do_dead_code_unlinked(exec_list *instructions);
bool opt_flip_matrices(exec_list *instructions);
bool do_lower_jumps(exec_list *instructions, bool pull_out_jumps = true, bool lower_continue = false);
bool do_if_simplification(exec_list *instructions);
bool opt_flatten_nested_if_blocks(exec_list *instructions);
src/compiler/glsl/meson.build
-1
View File
@@ -179,7 +179,6 @@ files_libglsl = files(
'opt_dead_builtin_variables.cpp',
'opt_dead_code.cpp',
'opt_flatten_nested_if_blocks.cpp',
'opt_flip_matrices.cpp',
'opt_function_inlining.cpp',
'opt_if_simplification.cpp',
'opt_minmax.cpp',
src/compiler/glsl/opt_flip_matrices.cpp
-123
View File
@@ -1,123 +0,0 @@
/*
* Copyright © 2013 Intel Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
* DEALINGS IN THE SOFTWARE.
*/
/**
* \file opt_flip_matrices.cpp
*
* Convert (matrix * vector) operations to (vector * matrixTranspose),
* which can be done using dot products rather than multiplies and adds.
* On some hardware, this is more efficient.
*
* This currently only does the conversion for built-in matrices which
* already have transposed equivalents. Namely, gl_ModelViewProjectionMatrix
* and gl_TextureMatrix.
*/
#include "ir.h"
#include "ir_optimization.h"
#include "main/macros.h"
namespace {
class matrix_flipper : public ir_hierarchical_visitor {
public:
matrix_flipper(exec_list *instructions)
{
progress = false;
mvp_transpose = NULL;
texmat_transpose = NULL;
foreach_in_list(ir_instruction, ir, instructions) {
ir_variable *var = ir->as_variable();
if (!var)
continue;
if (strcmp(var->name, "gl_ModelViewProjectionMatrixTranspose") == 0)
mvp_transpose = var;
if (strcmp(var->name, "gl_TextureMatrixTranspose") == 0)
texmat_transpose = var;
}
}
ir_visitor_status visit_enter(ir_expression *ir);
bool progress;
private:
ir_variable *mvp_transpose;
ir_variable *texmat_transpose;
};
}
ir_visitor_status
matrix_flipper::visit_enter(ir_expression *ir)
{
if (ir->operation != ir_binop_mul ||
!glsl_type_is_matrix(ir->operands[0]->type) ||
!glsl_type_is_vector(ir->operands[1]->type))
return visit_continue;
ir_variable *mat_var = ir->operands[0]->variable_referenced();
if (!mat_var)
return visit_continue;
if (mvp_transpose &&
strcmp(mat_var->name, "gl_ModelViewProjectionMatrix") == 0) {
#ifndef NDEBUG
ir_dereference_variable *deref = ir->operands[0]->as_dereference_variable();
assert(deref && deref->var == mat_var);
#endif
void *mem_ctx = ralloc_parent(ir);
ir->operands[0] = ir->operands[1];
ir->operands[1] = new(mem_ctx) ir_dereference_variable(mvp_transpose);
progress = true;
} else if (texmat_transpose &&
strcmp(mat_var->name, "gl_TextureMatrix") == 0) {
ir_dereference_array *array_ref = ir->operands[0]->as_dereference_array();
assert(array_ref != NULL);
ir_dereference_variable *var_ref = array_ref->array->as_dereference_variable();
assert(var_ref && var_ref->var == mat_var);
ir->operands[0] = ir->operands[1];
ir->operands[1] = array_ref;
var_ref->var = texmat_transpose;
texmat_transpose->data.max_array_access =
MAX2(texmat_transpose->data.max_array_access, mat_var->data.max_array_access);
progress = true;
}
return visit_continue;
}
bool
opt_flip_matrices(struct exec_list *instructions)
{
matrix_flipper v(instructions);
visit_list_elements(&v, instructions);
return v.progress;
}
src/mesa/main/consts_exts.h
-9
View File
@@ -353,15 +353,6 @@ struct gl_shader_compiler_options
GLuint MaxIfDepth; /**< Maximum nested IF blocks */
/**
* Optimize code for array of structures backends.
*
* This is a proxy for:
* - preferring DP4 instructions (rather than MUL/MAD) for
* matrix * vector operations, such as position transformation.
*/
GLboolean OptimizeForAOS;
/** Clamp UBO and SSBO block indices so they don't go out-of-bounds. */
GLboolean ClampBlockIndicesToArrayBounds;
src/mesa/main/ffvertex_prog.c
+11 -28
View File
@@ -281,7 +281,6 @@ static void make_state_key( struct gl_context *ctx, struct state_key *key )
struct tnl_program {
const struct state_key *state;
struct gl_program_parameter_list *state_params;
GLboolean mvp_with_dp4;
nir_builder *b;
@@ -457,18 +456,11 @@ get_eye_position(struct tnl_program *p)
if (!p->eye_position) {
nir_def *pos =
load_input_vec4(p, VERT_ATTRIB_POS);
if (p->mvp_with_dp4) {
nir_def *modelview[4];
load_state_mat4(p, modelview, STATE_MODELVIEW_MATRIX, 0);
p->eye_position =
emit_matrix_transform_vec4(p->b, modelview, pos);
} else {
nir_def *modelview[4];
load_state_mat4(p, modelview,
STATE_MODELVIEW_MATRIX_TRANSPOSE, 0);
p->eye_position =
emit_transpose_matrix_transform_vec4(p->b, modelview, pos);
}
nir_def *modelview[4];
load_state_mat4(p, modelview,
STATE_MODELVIEW_MATRIX_TRANSPOSE, 0);
p->eye_position =
emit_transpose_matrix_transform_vec4(p->b, modelview, pos);
}
return p->eye_position;
@@ -1216,17 +1208,11 @@ static void build_texture_transform( struct tnl_program *p )
if (p->state->unit[i].texmat_enabled) {
nir_def *texmat[4];
if (p->mvp_with_dp4) {
load_state_mat4(p, texmat, STATE_TEXTURE_MATRIX, i);
texcoord =
emit_matrix_transform_vec4(p->b, texmat, texcoord);
} else {
load_state_mat4(p, texmat,
STATE_TEXTURE_MATRIX_TRANSPOSE, i);
texcoord =
emit_transpose_matrix_transform_vec4(p->b, texmat,
texcoord);
}
load_state_mat4(p, texmat,
STATE_TEXTURE_MATRIX_TRANSPOSE, i);
texcoord =
emit_transpose_matrix_transform_vec4(p->b, texmat,
texcoord);
}
store_output_vec4(p, VARYING_SLOT_TEX0 + i, texcoord);
@@ -1321,14 +1307,12 @@ static void build_tnl_program( struct tnl_program *p )
static nir_shader *
create_new_program( const struct state_key *key,
struct gl_program *program,
GLboolean mvp_with_dp4,
const nir_shader_compiler_options *options)
{
struct tnl_program p;
memset(&p, 0, sizeof(p));
p.state = key;
p.mvp_with_dp4 = mvp_with_dp4;
program->Parameters = _mesa_new_parameter_list();
p.state_params = _mesa_new_parameter_list();
@@ -1349,7 +1333,7 @@ create_new_program( const struct state_key *key,
nir_validate_shader(b.shader, "after generating ff-vertex shader");
/* Emit the MVP position transformation */
NIR_PASS(_, b.shader, st_nir_lower_position_invariant, mvp_with_dp4, p.state_params);
NIR_PASS(_, b.shader, st_nir_lower_position_invariant, p.state_params);
_mesa_add_separate_state_parameters(program, p.state_params);
_mesa_free_parameter_list(p.state_params);
@@ -1394,7 +1378,6 @@ _mesa_get_fixed_func_vertex_program(struct gl_context *ctx)
nir_shader *s =
create_new_program( &key, prog,
ctx->Const.ShaderCompilerOptions[MESA_SHADER_VERTEX].OptimizeForAOS,
options);
prog->state.type = PIPE_SHADER_IR_NIR;
src/mesa/program/prog_to_nir.c
+1 -3
View File
@@ -857,9 +857,7 @@ prog_to_nir(const struct gl_context *ctx, const struct gl_program *prog)
/* ARB_vp: */
if (prog->arb.IsPositionInvariant) {
NIR_PASS(_, s, st_nir_lower_position_invariant,
ctx->Const.ShaderCompilerOptions[MESA_SHADER_VERTEX].OptimizeForAOS,
prog->Parameters);
NIR_PASS(_, s, st_nir_lower_position_invariant, prog->Parameters);
}
/* Add OPTION ARB_fog_exp code */
src/mesa/state_tracker/st_context.c
-8
View File
@@ -71,8 +71,6 @@
#include "compiler/glsl/glsl_parser_extras.h"
#include "nir.h"
DEBUG_GET_ONCE_BOOL_OPTION(mesa_mvp_dp4, "MESA_MVP_DP4", false)
void
st_invalidate_buffers(struct st_context *st)
{
@@ -833,12 +831,6 @@ st_create_context(gl_api api, struct pipe_context *pipe,
if (pipe->screen->get_disk_shader_cache)
ctx->Cache = pipe->screen->get_disk_shader_cache(pipe->screen);
/* XXX: need a capability bit in gallium to query if the pipe
* driver prefers DP4 or MUL/MAD for vertex transformation.
*/
if (debug_get_option_mesa_mvp_dp4())
ctx->Const.ShaderCompilerOptions[MESA_SHADER_VERTEX].OptimizeForAOS = GL_TRUE;
if (pipe->screen->caps.invalidate_buffer)
ctx->has_invalidate_buffer = true;
src/mesa/state_tracker/st_nir.h
+1 -1
View File
@@ -77,7 +77,7 @@ st_nir_state_variable_create(struct nir_shader *shader,
bool st_nir_lower_fog(struct nir_shader *s, enum gl_fog_mode fog_mode,
struct gl_program_parameter_list *paramList);
bool st_nir_lower_position_invariant(struct nir_shader *s, bool aos,
bool st_nir_lower_position_invariant(struct nir_shader *s,
struct gl_program_parameter_list *paramList);
bool st_nir_unlower_io_to_vars(struct nir_shader *nir);
src/mesa/state_tracker/st_nir_lower_position_invariant.c
+5 -12
View File
@@ -17,7 +17,7 @@
* both FF VS and ARB_vp.
*/
bool
st_nir_lower_position_invariant(struct nir_shader *s, bool aos,
st_nir_lower_position_invariant(struct nir_shader *s,
struct gl_program_parameter_list *paramList)
{
assert(s->info.io_lowered);
@@ -27,7 +27,7 @@ st_nir_lower_position_invariant(struct nir_shader *s, bool aos,
nir_def *mvp[4];
for (int i = 0; i < 4; i++) {
gl_state_index16 tokens[STATE_LENGTH] = {
aos ? STATE_MVP_MATRIX : STATE_MVP_MATRIX_TRANSPOSE, 0, i, i};
STATE_MVP_MATRIX_TRANSPOSE, 0, i, i};
nir_variable *var = st_nir_state_variable_create(s, glsl_vec4_type(), tokens);
_mesa_add_state_reference(paramList, tokens);
mvp[i] = nir_load_var(&b, var);
@@ -37,16 +37,9 @@ st_nir_lower_position_invariant(struct nir_shader *s, bool aos,
nir_def *in_pos = nir_load_input(&b, 4, 32, nir_imm_int(&b, 0),
.io_semantics.location = VERT_ATTRIB_POS);
if (aos) {
nir_def *chans[4];
for (int i = 0; i < 4; i++)
chans[i] = nir_fdot4(&b, mvp[i], in_pos);
result = nir_vec4(&b, chans[0], chans[1], chans[2], chans[3]);
} else {
result = nir_fmul(&b, mvp[0], nir_channel(&b, in_pos, 0));
for (int i = 1; i < 4; i++)
result = nir_fmad(&b, mvp[i], nir_channel(&b, in_pos, i), result);
}
result = nir_fmul(&b, mvp[0], nir_channel(&b, in_pos, 0));
for (int i = 1; i < 4; i++)
result = nir_fmad(&b, mvp[i], nir_channel(&b, in_pos, i), result);
nir_store_output(&b, result, nir_imm_int(&b, 0),
.io_semantics.location = VARYING_SLOT_POS);