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 FDIV_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 #define DDIV_TO_MUL_RCP 0x100000 53 #define DIV_TO_MUL_RCP (FDIV_TO_MUL_RCP | DDIV_TO_MUL_RCP) 54 55 /** 56 * \see class lower_packing_builtins_visitor 57 */ 58 enum lower_packing_builtins_op { 59 LOWER_PACK_UNPACK_NONE = 0x0000, 60 61 LOWER_PACK_SNORM_2x16 = 0x0001, 62 LOWER_UNPACK_SNORM_2x16 = 0x0002, 63 64 LOWER_PACK_UNORM_2x16 = 0x0004, 65 LOWER_UNPACK_UNORM_2x16 = 0x0008, 66 67 LOWER_PACK_HALF_2x16 = 0x0010, 68 LOWER_UNPACK_HALF_2x16 = 0x0020, 69 70 LOWER_PACK_SNORM_4x8 = 0x0040, 71 LOWER_UNPACK_SNORM_4x8 = 0x0080, 72 73 LOWER_PACK_UNORM_4x8 = 0x0100, 74 LOWER_UNPACK_UNORM_4x8 = 0x0200, 75 76 LOWER_PACK_USE_BFI = 0x0400, 77 LOWER_PACK_USE_BFE = 0x0800, 78 }; 79 80 bool do_common_optimization(exec_list *ir, bool linked, 81 bool uniform_locations_assigned, 82 const struct gl_shader_compiler_options *options, 83 bool native_integers); 84 85 bool ir_constant_fold(ir_rvalue **rvalue); 86 87 bool do_rebalance_tree(exec_list *instructions); 88 bool do_algebraic(exec_list *instructions, bool native_integers, 89 const struct gl_shader_compiler_options *options); 90 bool opt_conditional_discard(exec_list *instructions); 91 bool do_constant_folding(exec_list *instructions); 92 bool do_constant_variable(exec_list *instructions); 93 bool do_constant_variable_unlinked(exec_list *instructions); 94 bool do_copy_propagation(exec_list *instructions); 95 bool do_copy_propagation_elements(exec_list *instructions); 96 bool do_constant_propagation(exec_list *instructions); 97 void do_dead_builtin_varyings(struct gl_context *ctx, 98 gl_linked_shader *producer, 99 gl_linked_shader *consumer, 100 unsigned num_tfeedback_decls, 101 class tfeedback_decl *tfeedback_decls); 102 bool do_dead_code(exec_list *instructions, bool uniform_locations_assigned); 103 bool do_dead_code_local(exec_list *instructions); 104 bool do_dead_code_unlinked(exec_list *instructions); 105 bool do_dead_functions(exec_list *instructions); 106 bool opt_flip_matrices(exec_list *instructions); 107 bool do_function_inlining(exec_list *instructions); 108 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); 109 bool do_lower_texture_projection(exec_list *instructions); 110 bool do_if_simplification(exec_list *instructions); 111 bool opt_flatten_nested_if_blocks(exec_list *instructions); 112 bool do_discard_simplification(exec_list *instructions); 113 bool lower_if_to_cond_assign(gl_shader_stage stage, exec_list *instructions, 114 unsigned max_depth = 0, unsigned min_branch_cost = 0); 115 bool do_mat_op_to_vec(exec_list *instructions); 116 bool do_minmax_prune(exec_list *instructions); 117 bool do_noop_swizzle(exec_list *instructions); 118 bool do_structure_splitting(exec_list *instructions); 119 bool do_swizzle_swizzle(exec_list *instructions); 120 bool do_vectorize(exec_list *instructions); 121 bool do_tree_grafting(exec_list *instructions); 122 bool do_vec_index_to_cond_assign(exec_list *instructions); 123 bool do_vec_index_to_swizzle(exec_list *instructions); 124 bool lower_discard(exec_list *instructions); 125 void lower_discard_flow(exec_list *instructions); 126 bool lower_instructions(exec_list *instructions, unsigned what_to_lower); 127 bool lower_noise(exec_list *instructions); 128 bool lower_variable_index_to_cond_assign(gl_shader_stage stage, 129 exec_list *instructions, bool lower_input, bool lower_output, 130 bool lower_temp, bool lower_uniform); 131 bool lower_quadop_vector(exec_list *instructions, bool dont_lower_swz); 132 bool lower_const_arrays_to_uniforms(exec_list *instructions, unsigned stage); 133 bool lower_clip_cull_distance(struct gl_shader_program *prog, 134 gl_linked_shader *shader); 135 void lower_output_reads(unsigned stage, exec_list *instructions); 136 bool lower_packing_builtins(exec_list *instructions, int op_mask); 137 void lower_shared_reference(struct gl_linked_shader *shader, 138 unsigned *shared_size); 139 void lower_ubo_reference(struct gl_linked_shader *shader, 140 bool clamp_block_indices); 141 void lower_packed_varyings(void *mem_ctx, 142 unsigned locations_used, 143 const uint8_t *components, 144 ir_variable_mode mode, 145 unsigned gs_input_vertices, 146 gl_linked_shader *shader, 147 bool disable_varying_packing, bool xfb_enabled); 148 bool lower_vector_insert(exec_list *instructions, bool lower_nonconstant_index); 149 bool lower_vector_derefs(gl_linked_shader *shader); 150 void lower_named_interface_blocks(void *mem_ctx, gl_linked_shader *shader); 151 bool optimize_redundant_jumps(exec_list *instructions); 152 bool optimize_split_arrays(exec_list *instructions, bool linked); 153 bool lower_offset_arrays(exec_list *instructions); 154 void optimize_dead_builtin_variables(exec_list *instructions, 155 enum ir_variable_mode other); 156 bool lower_tess_level(gl_linked_shader *shader); 157 158 bool lower_vertex_id(gl_linked_shader *shader); 159 bool lower_blend_equation_advanced(gl_linked_shader *shader); 160 161 bool lower_subroutine(exec_list *instructions, struct _mesa_glsl_parse_state *state); 162 void propagate_invariance(exec_list *instructions); 163 164 ir_rvalue * 165 compare_index_block(exec_list *instructions, ir_variable *index, 166 unsigned base, unsigned components, void *mem_ctx); 167
