src/compiler/glsl/ir_optimization.h

   1 /*
   2  * Copyright © 2010 Intel Corporation
   3  *
   4  * Permission is hereby granted, free of charge, to any person obtaining a
   5  * copy of this software and associated documentation files (the "Software"),
   6  * to deal in the Software without restriction, including without limitation
   7  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
   8  * and/or sell copies of the Software, and to permit persons to whom the
   9  * Software is furnished to do so, subject to the following conditions:
  10  *
  11  * The above copyright notice and this permission notice (including the next
  12  * paragraph) shall be included in all copies or substantial portions of the
  13  * Software.
  14  *
  15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  16  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
  18  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  19  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  20  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
  21  * DEALINGS IN THE SOFTWARE.
  22  */
  23
  24
  25 /**
  26  * \file ir_optimization.h
  27  *
  28  * Prototypes for optimization passes to be called by the compiler and drivers.
  29  */
  30
  31 /* Operations for lower_instructions() */
  32 #define SUB_TO_ADD_NEG     0x01
  33 #define DIV_TO_MUL_RCP     0x02
  34 #define EXP_TO_EXP2        0x04
  35 #define POW_TO_EXP2        0x08
  36 #define LOG_TO_LOG2        0x10
  37 #define MOD_TO_FLOOR       0x20
  38 #define INT_DIV_TO_MUL_RCP 0x40
  39 #define LDEXP_TO_ARITH     0x80
  40 #define CARRY_TO_ARITH     0x100
  41 #define BORROW_TO_ARITH    0x200
  42 #define SAT_TO_CLAMP       0x400
  43 #define DOPS_TO_DFRAC      0x800
  44 #define DFREXP_DLDEXP_TO_ARITH    0x1000
  45 #define BIT_COUNT_TO_MATH         0x02000
  46 #define EXTRACT_TO_SHIFTS         0x04000
  47 #define INSERT_TO_SHIFTS          0x08000
  48 #define REVERSE_TO_SHIFTS         0x10000
  49 #define FIND_LSB_TO_FLOAT_CAST    0x20000
  50 #define FIND_MSB_TO_FLOAT_CAST    0x40000
  51 #define IMUL_HIGH_TO_MUL          0x80000
  52
  53 /* Opertaions for lower_64bit_integer_instructions() */
  54 #define MUL64                     (1U << 0)
  55
  56 /**
  57  * \see class lower_packing_builtins_visitor
  58  */
  59 enum lower_packing_builtins_op {
  60    LOWER_PACK_UNPACK_NONE               = 0x0000,
  61
  62    LOWER_PACK_SNORM_2x16                = 0x0001,
  63    LOWER_UNPACK_SNORM_2x16              = 0x0002,
  64
  65    LOWER_PACK_UNORM_2x16                = 0x0004,
  66    LOWER_UNPACK_UNORM_2x16              = 0x0008,
  67
  68    LOWER_PACK_HALF_2x16                 = 0x0010,
  69    LOWER_UNPACK_HALF_2x16               = 0x0020,
  70
  71    LOWER_PACK_SNORM_4x8                 = 0x0040,
  72    LOWER_UNPACK_SNORM_4x8               = 0x0080,
  73
  74    LOWER_PACK_UNORM_4x8                 = 0x0100,
  75    LOWER_UNPACK_UNORM_4x8               = 0x0200,
  76
  77    LOWER_PACK_USE_BFI                   = 0x0400,
  78    LOWER_PACK_USE_BFE                   = 0x0800,
  79 };
  80
  81 bool do_common_optimization(exec_list *ir, bool linked,
  82                             bool uniform_locations_assigned,
  83                             const struct gl_shader_compiler_options *options,
  84                             bool native_integers);
  85
  86 bool ir_constant_fold(ir_rvalue **rvalue);
  87
  88 bool do_rebalance_tree(exec_list *instructions);
  89 bool do_algebraic(exec_list *instructions, bool native_integers,
  90                   const struct gl_shader_compiler_options *options);
  91 bool opt_conditional_discard(exec_list *instructions);
  92 bool do_constant_folding(exec_list *instructions);
  93 bool do_constant_variable(exec_list *instructions);
  94 bool do_constant_variable_unlinked(exec_list *instructions);
  95 bool do_copy_propagation(exec_list *instructions);
  96 bool do_copy_propagation_elements(exec_list *instructions);
  97 bool do_constant_propagation(exec_list *instructions);
  98 void do_dead_builtin_varyings(struct gl_context *ctx,
  99                               gl_linked_shader *producer,
 100                               gl_linked_shader *consumer,
 101                               unsigned num_tfeedback_decls,
 102                               class tfeedback_decl *tfeedback_decls);
 103 bool do_dead_code(exec_list *instructions, bool uniform_locations_assigned);
 104 bool do_dead_code_local(exec_list *instructions);
 105 bool do_dead_code_unlinked(exec_list *instructions);
 106 bool do_dead_functions(exec_list *instructions);
 107 bool opt_flip_matrices(exec_list *instructions);
 108 bool do_function_inlining(exec_list *instructions);
 109 bool do_lower_jumps(exec_list *instructions, bool pull_out_jumps = true, bool lower_sub_return = true, bool lower_main_return = false, bool lower_continue = false, bool lower_break = false);
 110 bool do_lower_texture_projection(exec_list *instructions);
 111 bool do_if_simplification(exec_list *instructions);
 112 bool opt_flatten_nested_if_blocks(exec_list *instructions);
 113 bool do_discard_simplification(exec_list *instructions);
 114 bool lower_if_to_cond_assign(gl_shader_stage stage, exec_list *instructions,
 115                              unsigned max_depth = 0, unsigned min_branch_cost = 0);
 116 bool do_mat_op_to_vec(exec_list *instructions);
 117 bool do_minmax_prune(exec_list *instructions);
 118 bool do_noop_swizzle(exec_list *instructions);
 119 bool do_structure_splitting(exec_list *instructions);
 120 bool do_swizzle_swizzle(exec_list *instructions);
 121 bool do_vectorize(exec_list *instructions);
 122 bool do_tree_grafting(exec_list *instructions);
 123 bool do_vec_index_to_cond_assign(exec_list *instructions);
 124 bool do_vec_index_to_swizzle(exec_list *instructions);
 125 bool lower_discard(exec_list *instructions);
 126 void lower_discard_flow(exec_list *instructions);
 127 bool lower_instructions(exec_list *instructions, unsigned what_to_lower);
 128 bool lower_noise(exec_list *instructions);
 129 bool lower_variable_index_to_cond_assign(gl_shader_stage stage,
 130     exec_list *instructions, bool lower_input, bool lower_output,
 131     bool lower_temp, bool lower_uniform);
 132 bool lower_quadop_vector(exec_list *instructions, bool dont_lower_swz);
 133 bool lower_const_arrays_to_uniforms(exec_list *instructions, unsigned stage);
 134 bool lower_clip_cull_distance(struct gl_shader_program *prog,
 135                               gl_linked_shader *shader);
 136 void lower_output_reads(unsigned stage, exec_list *instructions);
 137 bool lower_packing_builtins(exec_list *instructions, int op_mask);
 138 void lower_shared_reference(struct gl_linked_shader *shader,
 139                             unsigned *shared_size);
 140 void lower_ubo_reference(struct gl_linked_shader *shader,
 141                          bool clamp_block_indices);
 142 void lower_packed_varyings(void *mem_ctx,
 143                            unsigned locations_used,
 144                            const uint8_t *components,
 145                            ir_variable_mode mode,
 146                            unsigned gs_input_vertices,
 147                            gl_linked_shader *shader,
 148                            bool disable_varying_packing, bool xfb_enabled);
 149 bool lower_vector_insert(exec_list *instructions, bool lower_nonconstant_index);
 150 bool lower_vector_derefs(gl_linked_shader *shader);
 151 void lower_named_interface_blocks(void *mem_ctx, gl_linked_shader *shader);
 152 bool optimize_redundant_jumps(exec_list *instructions);
 153 bool optimize_split_arrays(exec_list *instructions, bool linked);
 154 bool lower_offset_arrays(exec_list *instructions);
 155 void optimize_dead_builtin_variables(exec_list *instructions,
 156                                      enum ir_variable_mode other);
 157 bool lower_tess_level(gl_linked_shader *shader);
 158
 159 bool lower_vertex_id(gl_linked_shader *shader);
 160 bool lower_blend_equation_advanced(gl_linked_shader *shader);
 161
 162 bool lower_subroutine(exec_list *instructions, struct _mesa_glsl_parse_state *state);
 163 void propagate_invariance(exec_list *instructions);
 164
 165 ir_rvalue *
 166 compare_index_block(exec_list *instructions, ir_variable *index,
 167                     unsigned base, unsigned components, void *mem_ctx);
 168
 169 bool lower_64bit_integer_instructions(exec_list *instructions,
 170                                       unsigned what_to_lower);