diff --git a/.pick_status.json b/.pick_status.json
index d1785690af7..5e34f402f6c 100644
--- a/.pick_status.json
+++ b/.pick_status.json
@@ -1687,7 +1687,7 @@
         "description": "aco: Don't use opsel for p_insert.",
         "nominated": true,
         "nomination_type": 1,
-        "resolution": 0,
+        "resolution": 1,
         "main_sha": null,
         "because_sha": "54292e99c7844500314bfd623469c65adef954c5"
     },
diff --git a/src/amd/compiler/aco_optimizer.cpp b/src/amd/compiler/aco_optimizer.cpp
index a3f8555e15e..94fbfabded4 100644
--- a/src/amd/compiler/aco_optimizer.cpp
+++ b/src/amd/compiler/aco_optimizer.cpp
@@ -3245,21 +3245,13 @@ apply_insert(opt_ctx& ctx, aco_ptr<Instruction>& instr)
    SubdwordSel sel = parse_insert(def_info.instr);
    assert(sel);
 
-   if (instr->isVOP3() && sel.size() == 2 && !sel.sign_extend() &&
-       can_use_opsel(ctx.program->gfx_level, instr->opcode, -1)) {
-      if (instr->vop3().opsel & (1 << 3))
-         return false;
-      if (sel.offset())
-         instr->vop3().opsel |= 1 << 3;
-   } else {
-      if (!can_use_SDWA(ctx.program->gfx_level, instr, true))
-         return false;
+   if (!can_use_SDWA(ctx.program->gfx_level, instr, true))
+      return false;
 
-      to_SDWA(ctx, instr);
-      if (instr->sdwa().dst_sel.size() != 4)
-         return false;
-      static_cast<SDWA_instruction*>(instr.get())->dst_sel = sel;
-   }
+   to_SDWA(ctx, instr);
+   if (instr->sdwa().dst_sel.size() != 4)
+      return false;
+   static_cast<SDWA_instruction*>(instr.get())->dst_sel = sel;
 
    instr->definitions[0].swapTemp(def_info.instr->definitions[0]);
    ctx.info[instr->definitions[0].tempId()].label = 0;
diff --git a/src/amd/compiler/tests/test_sdwa.cpp b/src/amd/compiler/tests/test_sdwa.cpp
index 8d784ef6883..9df87eb568c 100644
--- a/src/amd/compiler/tests/test_sdwa.cpp
+++ b/src/amd/compiler/tests/test_sdwa.cpp
@@ -496,17 +496,15 @@ BEGIN_TEST(optimize.sdwa.insert)
       bld.pseudo(ins, bld.def(v1), val, Operand::c32(1u), Operand::c32(16u));
       writeout(10, val);
 
-      //~gfx8! v1: %tmp11 = v_sub_i16 %a, %b
-      //~gfx8! v1: %res11 = p_insert %tmp11, 0, 16
-      //~gfx(9|10)! v1: %res11 = v_sub_i16 %a, %b
-      //~gfx(8|9|10)! p_unit_test 11, %res11
+      //~gfx[^7]! v1: %tmp11 = v_sub_i16 %a, %b
+      //~gfx[^7]! v1: %res11 = p_insert %tmp11, 0, 16
+      //~gfx[^7]! p_unit_test 11, %res11
       val = bld.vop3(aco_opcode::v_sub_i16, bld.def(v1), inputs[0], inputs[1]);
       writeout(11, bld.pseudo(ins, bld.def(v1), val, Operand::zero(), Operand::c32(16u)));
 
-      //~gfx8! v1: %tmp12 = v_sub_i16 %a, %b
-      //~gfx8! v1: %res12 = p_insert %tmp12, 1, 16
-      //~gfx(9|10)! v1: %res12 = v_sub_i16 %a, %b opsel_hi
-      //~gfx(8|9|10)! p_unit_test 12, %res12
+      //~gfx[^7]! v1: %tmp12 = v_sub_i16 %a, %b
+      //~gfx[^7]! v1: %res12 = p_insert %tmp12, 1, 16
+      //~gfx[^7]! p_unit_test 12, %res12
       val = bld.vop3(aco_opcode::v_sub_i16, bld.def(v1), inputs[0], inputs[1]);
       writeout(12, bld.pseudo(ins, bld.def(v1), val, Operand::c32(1u), Operand::c32(16u)));