src/mesa/drivers/dri/i965/brw_vec4_cmod_propagation.cpp

   1 /*
   2  * Copyright © 2015 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 DEALINGS
  21  * IN THE SOFTWARE.
  22  *
  23  */
  24
  25 /** @file brw_vec4_cmod_propagation.cpp
  26  *
  27  * Really similar to brw_fs_cmod_propagation but adapted to vec4 needs. Check
  28  * brw_fs_cmod_propagation for further details on the rationale behind this
  29  * optimization.
  30  */
  31
  32 #include "brw_vec4.h"
  33 #include "brw_cfg.h"
  34 #include "brw_eu.h"
  35
  36 namespace brw {
  37
  38 static bool
  39 opt_cmod_propagation_local(bblock_t *block)
  40 {
  41    bool progress = false;
  42    int ip = block->end_ip + 1;
  43
  44    foreach_inst_in_block_reverse_safe(vec4_instruction, inst, block) {
  45       ip--;
  46
  47       if ((inst->opcode != BRW_OPCODE_AND &&
  48            inst->opcode != BRW_OPCODE_CMP &&
  49            inst->opcode != BRW_OPCODE_MOV) ||
  50           inst->predicate != BRW_PREDICATE_NONE ||
  51           !inst->dst.is_null() ||
  52           inst->src[0].file != VGRF ||
  53           inst->src[0].abs)
  54          continue;
  55
  56       if (inst->opcode == BRW_OPCODE_AND &&
  57           !(inst->src[1].is_one() &&
  58             inst->conditional_mod == BRW_CONDITIONAL_NZ &&
  59             !inst->src[0].negate))
  60          continue;
  61
  62       if (inst->opcode == BRW_OPCODE_CMP && !inst->src[1].is_zero())
  63          continue;
  64
  65       if (inst->opcode == BRW_OPCODE_MOV &&
  66           inst->conditional_mod != BRW_CONDITIONAL_NZ)
  67          continue;
  68
  69       bool read_flag = false;
  70       foreach_inst_in_block_reverse_starting_from(vec4_instruction, scan_inst, inst) {
  71          if (inst->src[0].in_range(scan_inst->dst,
  72                                    DIV_ROUND_UP(scan_inst->size_written, REG_SIZE))) {
  73             if ((scan_inst->predicate && scan_inst->opcode != BRW_OPCODE_SEL) ||
  74                 scan_inst->dst.offset / REG_SIZE != inst->src[0].offset / REG_SIZE ||
  75                 (scan_inst->dst.writemask != WRITEMASK_X &&
  76                  scan_inst->dst.writemask != WRITEMASK_XYZW) ||
  77                 (scan_inst->dst.writemask == WRITEMASK_XYZW &&
  78                  inst->src[0].swizzle != BRW_SWIZZLE_XYZW) ||
  79                 (inst->dst.writemask & ~scan_inst->dst.writemask) != 0) {
  80                break;
  81             }
  82
  83             /* CMP's result is the same regardless of dest type. */
  84             if (inst->conditional_mod == BRW_CONDITIONAL_NZ &&
  85                 scan_inst->opcode == BRW_OPCODE_CMP &&
  86                 (inst->dst.type == BRW_REGISTER_TYPE_D ||
  87                  inst->dst.type == BRW_REGISTER_TYPE_UD)) {
  88                inst->remove(block);
  89                progress = true;
  90                break;
  91             }
  92
  93             /* If the AND wasn't handled by the previous case, it isn't safe
  94              * to remove it.
  95              */
  96             if (inst->opcode == BRW_OPCODE_AND)
  97                break;
  98
  99             /* Comparisons operate differently for ints and floats */
 100             if (scan_inst->dst.type != inst->dst.type &&
 101                 (scan_inst->dst.type == BRW_REGISTER_TYPE_F ||
 102                  inst->dst.type == BRW_REGISTER_TYPE_F))
 103                break;
 104
 105             /* If the instruction generating inst's source also wrote the
 106              * flag, and inst is doing a simple .nz comparison, then inst
 107              * is redundant - the appropriate value is already in the flag
 108              * register.  Delete inst.
 109              */
 110             if (inst->conditional_mod == BRW_CONDITIONAL_NZ &&
 111                 !inst->src[0].negate &&
 112                 scan_inst->writes_flag()) {
 113                inst->remove(block);
 114                progress = true;
 115                break;
 116             }
 117
 118             /* The conditional mod of the CMP/CMPN instructions behaves
 119              * specially because the flag output is not calculated from the
 120              * result of the instruction, but the other way around, which
 121              * means that even if the condmod to propagate and the condmod
 122              * from the CMP instruction are the same they will in general give
 123              * different results because they are evaluated based on different
 124              * inputs.
 125              */
 126             if (scan_inst->opcode == BRW_OPCODE_CMP ||
 127                 scan_inst->opcode == BRW_OPCODE_CMPN)
 128                break;
 129
 130             /* Otherwise, try propagating the conditional. */
 131             enum brw_conditional_mod cond =
 132                inst->src[0].negate ? brw_swap_cmod(inst->conditional_mod)
 133                                    : inst->conditional_mod;
 134
 135             if (scan_inst->can_do_cmod() &&
 136                 ((!read_flag && scan_inst->conditional_mod == BRW_CONDITIONAL_NONE) ||
 137                  scan_inst->conditional_mod == cond)) {
 138                scan_inst->conditional_mod = cond;
 139                inst->remove(block);
 140                progress = true;
 141             }
 142             break;
 143          }
 144
 145          if (scan_inst->writes_flag())
 146             break;
 147
 148          read_flag = read_flag || scan_inst->reads_flag();
 149       }
 150    }
 151
 152    return progress;
 153 }
 154
 155 bool
 156 vec4_visitor::opt_cmod_propagation()
 157 {
 158    bool progress = false;
 159
 160    foreach_block_reverse(block, cfg) {
 161       progress = opt_cmod_propagation_local(block) || progress;
 162    }
 163
 164    if (progress)
 165       invalidate_live_intervals();
 166
 167    return progress;
 168 }
 169
 170 } /* namespace brw */