Home | History | Annotate | Download | only in glsl
      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  * \file ir_optimization.h
     26  *
     27  * Prototypes for optimization passes to be called by the compiler and drivers.
     28  */
     29 
     30 #ifndef GLSL_IR_OPTIMIZATION_H
     31 #define GLSL_IR_OPTIMIZATION_H
     32 
     33 /* Operations for lower_instructions() */
     34 #define SUB_TO_ADD_NEG     0x01
     35 #define FDIV_TO_MUL_RCP    0x02
     36 #define EXP_TO_EXP2        0x04
     37 #define POW_TO_EXP2        0x08
     38 #define LOG_TO_LOG2        0x10
     39 #define MOD_TO_FLOOR       0x20
     40 #define INT_DIV_TO_MUL_RCP 0x40
     41 #define LDEXP_TO_ARITH     0x80
     42 #define CARRY_TO_ARITH     0x100
     43 #define BORROW_TO_ARITH    0x200
     44 #define SAT_TO_CLAMP       0x400
     45 #define DOPS_TO_DFRAC      0x800
     46 #define DFREXP_DLDEXP_TO_ARITH    0x1000
     47 #define BIT_COUNT_TO_MATH         0x02000
     48 #define EXTRACT_TO_SHIFTS         0x04000
     49 #define INSERT_TO_SHIFTS          0x08000
     50 #define REVERSE_TO_SHIFTS         0x10000
     51 #define FIND_LSB_TO_FLOAT_CAST    0x20000
     52 #define FIND_MSB_TO_FLOAT_CAST    0x40000
     53 #define IMUL_HIGH_TO_MUL          0x80000
     54 #define DDIV_TO_MUL_RCP           0x100000
     55 #define DIV_TO_MUL_RCP            (FDIV_TO_MUL_RCP | DDIV_TO_MUL_RCP)
     56 #define SQRT_TO_ABS_SQRT          0x200000
     57 
     58 /* Opertaions for lower_64bit_integer_instructions() */
     59 #define MUL64                     (1U << 0)
     60 #define SIGN64                    (1U << 1)
     61 #define DIV64                     (1U << 2)
     62 #define MOD64                     (1U << 3)
     63 
     64 /**
     65  * \see class lower_packing_builtins_visitor
     66  */
     67 enum lower_packing_builtins_op {
     68    LOWER_PACK_UNPACK_NONE               = 0x0000,
     69 
     70    LOWER_PACK_SNORM_2x16                = 0x0001,
     71    LOWER_UNPACK_SNORM_2x16              = 0x0002,
     72 
     73    LOWER_PACK_UNORM_2x16                = 0x0004,
     74    LOWER_UNPACK_UNORM_2x16              = 0x0008,
     75 
     76    LOWER_PACK_HALF_2x16                 = 0x0010,
     77    LOWER_UNPACK_HALF_2x16               = 0x0020,
     78 
     79    LOWER_PACK_SNORM_4x8                 = 0x0040,
     80    LOWER_UNPACK_SNORM_4x8               = 0x0080,
     81 
     82    LOWER_PACK_UNORM_4x8                 = 0x0100,
     83    LOWER_UNPACK_UNORM_4x8               = 0x0200,
     84 
     85    LOWER_PACK_USE_BFI                   = 0x0400,
     86    LOWER_PACK_USE_BFE                   = 0x0800,
     87 };
     88 
     89 bool do_common_optimization(exec_list *ir, bool linked,
     90 			    bool uniform_locations_assigned,
     91                             const struct gl_shader_compiler_options *options,
     92                             bool native_integers);
     93 
     94 bool ir_constant_fold(ir_rvalue **rvalue);
     95 
     96 bool do_rebalance_tree(exec_list *instructions);
     97 bool do_algebraic(exec_list *instructions, bool native_integers,
     98                   const struct gl_shader_compiler_options *options);
     99 bool opt_conditional_discard(exec_list *instructions);
    100 bool do_constant_folding(exec_list *instructions);
    101 bool do_constant_variable(exec_list *instructions);
    102 bool do_constant_variable_unlinked(exec_list *instructions);
    103 bool do_copy_propagation(exec_list *instructions);
    104 bool do_copy_propagation_elements(exec_list *instructions);
    105 bool do_constant_propagation(exec_list *instructions);
    106 void do_dead_builtin_varyings(struct gl_context *ctx,
    107                               gl_linked_shader *producer,
    108                               gl_linked_shader *consumer,
    109                               unsigned num_tfeedback_decls,
    110                               class tfeedback_decl *tfeedback_decls);
    111 bool do_dead_code(exec_list *instructions, bool uniform_locations_assigned);
    112 bool do_dead_code_local(exec_list *instructions);
    113 bool do_dead_code_unlinked(exec_list *instructions);
    114 bool do_dead_functions(exec_list *instructions);
    115 bool opt_flip_matrices(exec_list *instructions);
    116 bool do_function_inlining(exec_list *instructions);
    117 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);
    118 bool do_lower_texture_projection(exec_list *instructions);
    119 bool do_if_simplification(exec_list *instructions);
    120 bool opt_flatten_nested_if_blocks(exec_list *instructions);
    121 bool do_discard_simplification(exec_list *instructions);
    122 bool lower_if_to_cond_assign(gl_shader_stage stage, exec_list *instructions,
    123                              unsigned max_depth = 0, unsigned min_branch_cost = 0);
    124 bool do_mat_op_to_vec(exec_list *instructions);
    125 bool do_minmax_prune(exec_list *instructions);
    126 bool do_structure_splitting(exec_list *instructions);
    127 bool optimize_swizzles(exec_list *instructions);
    128 bool do_vectorize(exec_list *instructions);
    129 bool do_tree_grafting(exec_list *instructions);
    130 bool do_vec_index_to_cond_assign(exec_list *instructions);
    131 bool do_vec_index_to_swizzle(exec_list *instructions);
    132 bool lower_discard(exec_list *instructions);
    133 void lower_discard_flow(exec_list *instructions);
    134 bool lower_instructions(exec_list *instructions, unsigned what_to_lower);
    135 bool lower_noise(exec_list *instructions);
    136 bool lower_variable_index_to_cond_assign(gl_shader_stage stage,
    137     exec_list *instructions, bool lower_input, bool lower_output,
    138     bool lower_temp, bool lower_uniform);
    139 bool lower_quadop_vector(exec_list *instructions, bool dont_lower_swz);
    140 bool lower_const_arrays_to_uniforms(exec_list *instructions, unsigned stage);
    141 bool lower_clip_cull_distance(struct gl_shader_program *prog,
    142                               gl_linked_shader *shader);
    143 void lower_output_reads(unsigned stage, exec_list *instructions);
    144 bool lower_packing_builtins(exec_list *instructions, int op_mask);
    145 void lower_shared_reference(struct gl_context *ctx,
    146                             struct gl_shader_program *prog,
    147                             struct gl_linked_shader *shader);
    148 void lower_ubo_reference(struct gl_linked_shader *shader,
    149                          bool clamp_block_indices, bool use_std430_as_default);
    150 void lower_packed_varyings(void *mem_ctx,
    151                            unsigned locations_used,
    152                            const uint8_t *components,
    153                            ir_variable_mode mode,
    154                            unsigned gs_input_vertices,
    155                            gl_linked_shader *shader,
    156                            bool disable_varying_packing, bool xfb_enabled);
    157 bool lower_vector_insert(exec_list *instructions, bool lower_nonconstant_index);
    158 bool lower_vector_derefs(gl_linked_shader *shader);
    159 void lower_named_interface_blocks(void *mem_ctx, gl_linked_shader *shader);
    160 bool optimize_redundant_jumps(exec_list *instructions);
    161 bool optimize_split_arrays(exec_list *instructions, bool linked);
    162 bool lower_offset_arrays(exec_list *instructions);
    163 void optimize_dead_builtin_variables(exec_list *instructions,
    164                                      enum ir_variable_mode other);
    165 bool lower_tess_level(gl_linked_shader *shader);
    166 
    167 bool lower_vertex_id(gl_linked_shader *shader);
    168 bool lower_cs_derived(gl_linked_shader *shader);
    169 bool lower_blend_equation_advanced(gl_linked_shader *shader);
    170 
    171 bool lower_subroutine(exec_list *instructions, struct _mesa_glsl_parse_state *state);
    172 void propagate_invariance(exec_list *instructions);
    173 
    174 namespace ir_builder { class ir_factory; };
    175 
    176 ir_variable *compare_index_block(ir_builder::ir_factory &body,
    177                                  ir_variable *index,
    178                                  unsigned base, unsigned components);
    179 
    180 bool lower_64bit_integer_instructions(exec_list *instructions,
    181                                       unsigned what_to_lower);
    182 
    183 #endif /* GLSL_IR_OPTIMIZATION_H */
    184