aco: use v_xor3_b32

fossil-db (Navi):
Totals from 334 (0.26% of 128321) affected shaders:
CodeSize: 3345532 -> 3345484 (-0.00%); split: -0.00%, +0.00%
Instrs: 624662 -> 622778 (-0.30%); split: -0.30%, +0.00%

Mostly affects some parallel-rdp shaders

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5357>

diff --git a/src/amd/compiler/aco_optimizer.cpp b/src/amd/compiler/aco_optimizer.cpp
index f16f165c21b765a7c1c3fa29c3f350db287f4b96..2c0bd59d29075b59e87bb76006fd49d3478b4f35 100644 (file)
--- a/src/amd/compiler/aco_optimizer.cpp
+++ b/src/amd/compiler/aco_optimizer.cpp
@@ -2542,6 +2542,9 @@ void combine_instruction(opt_ctx &ctx, Block& block, aco_ptr<Instruction>& instr
       else if (combine_three_valu_op(ctx, instr, aco_opcode::v_and_b32, aco_opcode::v_and_or_b32, "120", 1 | 2)) ;
       else if (combine_three_valu_op(ctx, instr, aco_opcode::s_lshl_b32, aco_opcode::v_lshl_or_b32, "120", 1 | 2)) ;
       else combine_three_valu_op(ctx, instr, aco_opcode::v_lshlrev_b32, aco_opcode::v_lshl_or_b32, "210", 1 | 2);
+   } else if (instr->opcode == aco_opcode::v_xor_b32 && ctx.program->chip_class >= GFX10) {
+      if (combine_three_valu_op(ctx, instr, aco_opcode::v_xor_b32, aco_opcode::v_xor3_b32, "012", 1 | 2)) ;
+      else combine_three_valu_op(ctx, instr, aco_opcode::s_xor_b32, aco_opcode::v_xor3_b32, "012", 1 | 2);
    } else if (instr->opcode == aco_opcode::v_add_u32) {
       if (combine_add_sub_b2i(ctx, instr, aco_opcode::v_addc_co_u32, 1 | 2)) ;
       else if (ctx.program->chip_class >= GFX9) {