diff --git a/src/amd/compiler/aco_optimizer.cpp b/src/amd/compiler/aco_optimizer.cpp
index 8492fcac6c1..a48988c0029 100644
--- a/src/amd/compiler/aco_optimizer.cpp
+++ b/src/amd/compiler/aco_optimizer.cpp
@@ -3694,7 +3694,8 @@ apply_output_impl(opt_ctx& ctx, aco_ptr<Instruction>& instr, Instruction* parent
    else if (instr->opcode == aco_opcode::v_mul_f64 || instr->opcode == aco_opcode::v_mul_f64_e64 ||
             instr->opcode == aco_opcode::v_mul_f32 || instr->opcode == aco_opcode::v_mul_f16 ||
             instr->opcode == aco_opcode::v_pk_mul_f16 ||
-            instr->opcode == aco_opcode::v_mul_legacy_f32)
+            instr->opcode == aco_opcode::v_mul_legacy_f32 ||
+            instr->opcode == aco_opcode::s_mul_f32 || instr->opcode == aco_opcode::s_mul_f16)
       return apply_output_mul(ctx, instr, parent);
    else if (instr->opcode == aco_opcode::v_cvt_f16_f32)
       return apply_f2f16(ctx, instr, parent);
@@ -3722,6 +3723,8 @@ apply_output(opt_ctx& ctx, aco_ptr<Instruction>& instr)
    case aco_opcode::v_mul_f16:
    case aco_opcode::v_pk_mul_f16:
    case aco_opcode::v_mul_legacy_f32:
+   case aco_opcode::s_mul_f32:
+   case aco_opcode::s_mul_f16:
    case aco_opcode::v_cvt_f16_f32:
    case aco_opcode::v_med3_f32:
    case aco_opcode::v_med3_f16: break;
diff --git a/src/amd/compiler/tests/test_optimizer.cpp b/src/amd/compiler/tests/test_optimizer.cpp
index 8abea38fc68..2906547f708 100644
--- a/src/amd/compiler/tests/test_optimizer.cpp
+++ b/src/amd/compiler/tests/test_optimizer.cpp
@@ -2040,7 +2040,7 @@ BEGIN_TEST(optimizer.trans_inline_constant)
    finish_opt_test();
 END_TEST
 
-BEGIN_TEST(optimizer.trans_no_omod)
+BEGIN_TEST(optimizer.trans_omod)
    //>> s1: %a:s[0] = p_startpgm
    if (!setup_cs("s1", GFX12))
       return;
@@ -2052,6 +2052,12 @@ BEGIN_TEST(optimizer.trans_no_omod)
    writeout(0, bld.vop2(aco_opcode::v_mul_legacy_f32, bld.def(v1), dst,
                         bld.copy(bld.def(v1), Operand::c32(0x3f000000))));
 
+   //! s1: %res1 = v_s_rcp_f32 -%a *0.5
+   //! p_unit_test 1, %res1
+   dst = bld.vop3(aco_opcode::v_s_rcp_f32, bld.def(s1), inputs[0]);
+   writeout(1, bld.sop2(aco_opcode::s_mul_f32, bld.def(s1), dst,
+                        bld.copy(bld.def(s1), Operand::c32(0xbf000000))));
+
    finish_opt_test();
 END_TEST