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
  35 namespace brw {
  36
  37 static bool
  38 opt_cmod_propagation_local(bblock_t *block)
  39 {
  40    bool progress = false;
  41    int ip = block->end_ip + 1;
  42
  43    foreach_inst_in_block_reverse_safe(vec4_instruction, inst, block) {
  44       ip--;
  45
  46       if ((inst->opcode != BRW_OPCODE_AND &&
  47            inst->opcode != BRW_OPCODE_CMP &&
  48            inst->opcode != BRW_OPCODE_MOV) ||
  49           inst->predicate != BRW_PREDICATE_NONE ||
  50           !inst->dst.is_null() ||
  51           inst->src[0].file != VGRF ||
  52           inst->src[0].abs)
  53          continue;
  54
  55       if (inst->opcode == BRW_OPCODE_AND &&
  56           !(inst->src[1].is_one() &&
  57             inst->conditional_mod == BRW_CONDITIONAL_NZ &&
  58             !inst->src[0].negate))
  59          continue;
  60
  61       if (inst->opcode == BRW_OPCODE_CMP && !inst->src[1].is_zero())
  62          continue;
  63
  64       if (inst->opcode == BRW_OPCODE_MOV &&
  65           inst->conditional_mod != BRW_CONDITIONAL_NZ)
  66          continue;
  67
  68       bool read_flag = false;
  69       foreach_inst_in_block_reverse_starting_from(vec4_instruction, scan_inst, inst) {
  70          if (inst->src[0].in_range(scan_inst->dst,
  71                                    scan_inst->regs_written)) {
  72             if ((scan_inst->predicate && scan_inst->opcode != BRW_OPCODE_SEL) ||
  73                 scan_inst->dst.reg_offset != inst->src[0].reg_offset ||
  74                 (scan_inst->dst.writemask != WRITEMASK_X &&
  75                  scan_inst->dst.writemask != WRITEMASK_XYZW) ||
  76                 (scan_inst->dst.writemask == WRITEMASK_XYZW &&
  77                  inst->src[0].swizzle != BRW_SWIZZLE_XYZW) ||
  78                 (inst->dst.writemask & ~scan_inst->dst.writemask) != 0) {
  79                break;
  80             }
  81
  82             /* CMP's result is the same regardless of dest type. */
  83             if (inst->conditional_mod == BRW_CONDITIONAL_NZ &&
  84                 scan_inst->opcode == BRW_OPCODE_CMP &&
  85                 (inst->dst.type == BRW_REGISTER_TYPE_D ||
  86                  inst->dst.type == BRW_REGISTER_TYPE_UD)) {
  87                inst->remove(block);
  88                progress = true;
  89                break;
  90             }
  91
  92             /* If the AND wasn't handled by the previous case, it isn't safe
  93              * to remove it.
  94              */
  95             if (inst->opcode == BRW_OPCODE_AND)
  96                break;
  97
  98             /* Comparisons operate differently for ints and floats */
  99             if (scan_inst->dst.type != inst->dst.type &&
 100                 (scan_inst->dst.type == BRW_REGISTER_TYPE_F ||
 101                  inst->dst.type == BRW_REGISTER_TYPE_F))
 102                break;
 103
 104             /* If the instruction generating inst's source also wrote the
 105              * flag, and inst is doing a simple .nz comparison, then inst
 106              * is redundant - the appropriate value is already in the flag
 107              * register.  Delete inst.
 108              */
 109             if (inst->conditional_mod == BRW_CONDITIONAL_NZ &&
 110                 !inst->src[0].negate &&
 111                 scan_inst->writes_flag()) {
 112                inst->remove(block);
 113                progress = true;
 114                break;
 115             }
 116
 117             /* Otherwise, try propagating the conditional. */
 118             enum brw_conditional_mod cond =
 119                inst->src[0].negate ? brw_swap_cmod(inst->conditional_mod)
 120                                    : inst->conditional_mod;
 121
 122             if (scan_inst->can_do_cmod() &&
 123                 ((!read_flag && scan_inst->conditional_mod == BRW_CONDITIONAL_NONE) ||
 124                  scan_inst->conditional_mod == cond)) {
 125                scan_inst->conditional_mod = cond;
 126                inst->remove(block);
 127                progress = true;
 128             }
 129             break;
 130          }
 131
 132          if (scan_inst->writes_flag())
 133             break;
 134
 135          read_flag = read_flag || scan_inst->reads_flag();
 136       }
 137    }
 138
 139    return progress;
 140 }
 141
 142 bool
 143 vec4_visitor::opt_cmod_propagation()
 144 {
 145    bool progress = false;
 146
 147    foreach_block_reverse(block, cfg) {
 148       progress = opt_cmod_propagation_local(block) || progress;
 149    }
 150
 151    if (progress)
 152       invalidate_live_intervals();
 153
 154    return progress;
 155 }
 156
 157 } /* namespace brw */