Mesa (master): aco: always set exec_live=false

[GitLab Mirror]

Module: Mesa
Branch: master
Commit: d1f93261b17d73eceb16b7081000b5bb8f6608f2
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=d1f93261b17d73eceb16b7081000b5bb8f6608f2

Author: Rhys Perry <pendingchaos02 at gmail.com>
Date:   Wed Jan 27 14:04:10 2021 +0000

aco: always set exec_live=false

Register demand calculation for exec masks doesn't always match
get_live_changes() and get_temp_registers(). For now, just set
exec_live=false.

fossil-db (GFX10.3):
Totals from 108230 (77.64% of 139391) affected shaders:
SGPRs: 5759658 -> 5756818 (-0.05%); split: -0.08%, +0.03%
VGPRs: 4061104 -> 4061248 (+0.00%); split: -0.00%, +0.01%
SpillSGPRs: 14114 -> 15198 (+7.68%); split: -0.10%, +7.78%
CodeSize: 266548396 -> 266603288 (+0.02%); split: -0.01%, +0.03%
MaxWaves: 1390885 -> 1390855 (-0.00%); split: +0.00%, -0.00%
Instrs: 50983353 -> 50992972 (+0.02%); split: -0.02%, +0.04%
Cycles: 1733042048 -> 1735443264 (+0.14%); split: -0.02%, +0.16%
VMEM: 41933625 -> 41914722 (-0.05%); split: +0.04%, -0.09%
SMEM: 7197675 -> 7197789 (+0.00%); split: +0.16%, -0.16%
VClause: 1050885 -> 1050978 (+0.01%); split: -0.02%, +0.03%
SClause: 2074913 -> 2071844 (-0.15%); split: -0.23%, +0.08%
Copies: 3181464 -> 3188125 (+0.21%); split: -0.38%, +0.59%
Branches: 1127526 -> 1127716 (+0.02%); split: -0.10%, +0.12%
PreSGPRs: 3376687 -> 3586076 (+6.20%); split: -0.00%, +6.20%
PreVGPRs: 3339740 -> 3339811 (+0.00%)

Signed-off-by: Rhys Perry <pendingchaos02 at gmail.com>
Reviewed-by: Daniel Schürmann <daniel at schuermann.dev>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8807>

---

 src/amd/compiler/aco_live_var_analysis.cpp | 21 +--------------------
 1 file changed, 1 insertion(+), 20 deletions(-)

diff --git a/src/amd/compiler/aco_live_var_analysis.cpp b/src/amd/compiler/aco_live_var_analysis.cpp
index 1c041c9dbbb..61a1c626c5c 100644
--- a/src/amd/compiler/aco_live_var_analysis.cpp
+++ b/src/amd/compiler/aco_live_var_analysis.cpp
@@ -91,13 +91,6 @@ void process_live_temps_per_block(Program *program, live& lives, Block* block,
    block->register_demand = RegisterDemand();
    IDSet live = lives.live_out[block->index];
 
-   /* add the live_out_exec to live */
-   bool exec_live = false;
-   if (block->live_out_exec != Temp()) {
-      live.insert(block->live_out_exec.id());
-      exec_live = true;
-   }
-
    /* initialize register demand */
    for (unsigned t : live)
       new_demand += Temp(t, program->temp_rc[t]);
@@ -110,10 +103,7 @@ void process_live_temps_per_block(Program *program, live& lives, Block* block,
       if (is_phi(insn))
          break;
 
-      /* substract the 1 or 2 sgprs from exec */
-      if (exec_live)
-         assert(new_demand.sgpr >= (int16_t) program->lane_mask.size());
-      register_demand[idx] = RegisterDemand(new_demand.vgpr, new_demand.sgpr - (exec_live ? program->lane_mask.size() : 0));
+      register_demand[idx] = RegisterDemand(new_demand.vgpr, new_demand.sgpr);
 
       /* KILL */
       for (Definition& definition : insn->definitions) {
@@ -133,9 +123,6 @@ void process_live_temps_per_block(Program *program, live& lives, Block* block,
             register_demand[idx] += temp;
             definition.setKill(true);
          }
-
-         if (definition.isFixed() && definition.physReg() == exec)
-            exec_live = false;
       }
 
       /* GEN */
@@ -169,9 +156,6 @@ void process_live_temps_per_block(Program *program, live& lives, Block* block,
                   register_demand[idx] += temp;
                new_demand += temp;
             }
-
-            if (operand.isFixed() && operand.physReg() == exec)
-               exec_live = true;
          }
       }
 
@@ -179,9 +163,6 @@ void process_live_temps_per_block(Program *program, live& lives, Block* block,
    }
 
    /* update block's register demand for a last time */
-   if (exec_live)
-      assert(new_demand.sgpr >= (int16_t) program->lane_mask.size());
-   new_demand.sgpr -= exec_live ? program->lane_mask.size() : 0;
    block->register_demand.update(new_demand);
 
    /* handle phi definitions */