diff --git a/src/amd/compiler/aco_assembler.cpp b/src/amd/compiler/aco_assembler.cpp
index f19f731b090..9da9dd30068 100644
--- a/src/amd/compiler/aco_assembler.cpp
+++ b/src/amd/compiler/aco_assembler.cpp
@@ -126,7 +126,7 @@ emit_instruction(asm_context& ctx, std::vector<uint32_t>& out, Instruction* inst
       FILE* const memf = u_memstream_get(&mem);
 
       fprintf(memf, "Unsupported opcode: ");
-      aco_print_instr(instr, memf);
+      aco_print_instr(ctx.gfx_level, instr, memf);
       u_memstream_close(&mem);
 
       aco_err(ctx.program, outmem);
diff --git a/src/amd/compiler/aco_ir.h b/src/amd/compiler/aco_ir.h
index 57e7834a997..fbdea652c66 100644
--- a/src/amd/compiler/aco_ir.h
+++ b/src/amd/compiler/aco_ir.h
@@ -2320,7 +2320,8 @@ enum print_flags {
 };
 
 void aco_print_operand(const Operand* operand, FILE* output, unsigned flags = 0);
-void aco_print_instr(const Instruction* instr, FILE* output, unsigned flags = 0);
+void aco_print_instr(enum amd_gfx_level gfx_level, const Instruction* instr, FILE* output,
+                     unsigned flags = 0);
 void aco_print_program(const Program* program, FILE* output, unsigned flags = 0);
 void aco_print_program(const Program* program, FILE* output, const live& live_vars,
                        unsigned flags = 0);
diff --git a/src/amd/compiler/aco_optimizer.cpp b/src/amd/compiler/aco_optimizer.cpp
index 945da8879ba..96d5eddf77e 100644
--- a/src/amd/compiler/aco_optimizer.cpp
+++ b/src/amd/compiler/aco_optimizer.cpp
@@ -46,7 +46,7 @@ perfwarn(Program* program, bool cond, const char* msg, Instruction* instr)
       FILE* const memf = u_memstream_get(&mem);
 
       fprintf(memf, "%s: ", msg);
-      aco_print_instr(instr, memf);
+      aco_print_instr(program->gfx_level, instr, memf);
       u_memstream_close(&mem);
 
       aco_perfwarn(program, out);
diff --git a/src/amd/compiler/aco_print_ir.cpp b/src/amd/compiler/aco_print_ir.cpp
index b45ad118590..dee0f891ca1 100644
--- a/src/amd/compiler/aco_print_ir.cpp
+++ b/src/amd/compiler/aco_print_ir.cpp
@@ -277,7 +277,7 @@ print_sync(memory_sync_info sync, FILE* output)
 }
 
 static void
-print_instr_format_specific(const Instruction* instr, FILE* output)
+print_instr_format_specific(enum amd_gfx_level gfx_level, const Instruction* instr, FILE* output)
 {
    switch (instr->format) {
    case Format::SOPK: {
@@ -289,15 +289,13 @@ print_instr_format_specific(const Instruction* instr, FILE* output)
       uint16_t imm = instr->sopp().imm;
       switch (instr->opcode) {
       case aco_opcode::s_waitcnt: {
-         /* we usually should check the gfx level for vmcnt/lgkm, but
-          * insert_waitcnt() should fill it in regardless. */
-         unsigned vmcnt = (imm & 0xF) | ((imm & (0x3 << 14)) >> 10);
-         if (vmcnt != 63)
-            fprintf(output, " vmcnt(%d)", vmcnt);
-         if (((imm >> 4) & 0x7) < 0x7)
-            fprintf(output, " expcnt(%d)", (imm >> 4) & 0x7);
-         if (((imm >> 8) & 0x3F) < 0x3F)
-            fprintf(output, " lgkmcnt(%d)", (imm >> 8) & 0x3F);
+         wait_imm unpacked(gfx_level, imm);
+         if (unpacked.vm != wait_imm::unset_counter)
+            fprintf(output, " vmcnt(%d)", unpacked.vm);
+         if (unpacked.exp != wait_imm::unset_counter)
+            fprintf(output, " expcnt(%d)", unpacked.exp);
+         if (unpacked.lgkm != wait_imm::unset_counter)
+            fprintf(output, " lgkmcnt(%d)", unpacked.lgkm);
          break;
       }
       case aco_opcode::s_waitcnt_depctr: {
@@ -700,7 +698,8 @@ print_instr_format_specific(const Instruction* instr, FILE* output)
 }
 
 void
-aco_print_instr(const Instruction* instr, FILE* output, unsigned flags)
+aco_print_instr(enum amd_gfx_level gfx_level, const Instruction* instr, FILE* output,
+                unsigned flags)
 {
    if (!instr->definitions.empty()) {
       for (unsigned i = 0; i < instr->definitions.size(); ++i) {
@@ -797,7 +796,7 @@ aco_print_instr(const Instruction* instr, FILE* output, unsigned flags)
          }
       }
    }
-   print_instr_format_specific(instr, output);
+   print_instr_format_specific(gfx_level, instr, output);
 }
 
 static void
@@ -883,7 +882,8 @@ print_stage(Stage stage, FILE* output)
 }
 
 void
-aco_print_block(const Block* block, FILE* output, unsigned flags, const live& live_vars)
+aco_print_block(enum amd_gfx_level gfx_level, const Block* block, FILE* output, unsigned flags,
+                const live& live_vars)
 {
    fprintf(output, "BB%d\n", block->index);
    fprintf(output, "/* logical preds: ");
@@ -916,7 +916,7 @@ aco_print_block(const Block* block, FILE* output, unsigned flags, const live& li
       if (flags & print_perf_info)
          fprintf(output, "(%3u clk)   ", instr->pass_flags);
 
-      aco_print_instr(instr.get(), output, flags);
+      aco_print_instr(gfx_level, instr.get(), output, flags);
       fprintf(output, "\n");
       index++;
    }
@@ -937,7 +937,7 @@ aco_print_program(const Program* program, FILE* output, const live& live_vars, u
    print_stage(program->stage, output);
 
    for (Block const& block : program->blocks)
-      aco_print_block(&block, output, flags, live_vars);
+      aco_print_block(program->gfx_level, &block, output, flags, live_vars);
 
    if (program->constant_data.size()) {
       fprintf(output, "\n/* constant data */\n");
diff --git a/src/amd/compiler/aco_validate.cpp b/src/amd/compiler/aco_validate.cpp
index abfb0b2d242..fef525df462 100644
--- a/src/amd/compiler/aco_validate.cpp
+++ b/src/amd/compiler/aco_validate.cpp
@@ -91,7 +91,7 @@ validate_ir(Program* program)
          FILE* const memf = u_memstream_get(&mem);
 
          fprintf(memf, "%s: ", msg);
-         aco_print_instr(instr, memf);
+         aco_print_instr(program->gfx_level, instr, memf);
          u_memstream_close(&mem);
 
          aco_err(program, "%s", out);
@@ -776,14 +776,14 @@ ra_fail(Program* program, Location loc, Location loc2, const char* fmt, ...)
 
    fprintf(memf, "RA error found at instruction in BB%d:\n", loc.block->index);
    if (loc.instr) {
-      aco_print_instr(loc.instr, memf);
+      aco_print_instr(program->gfx_level, loc.instr, memf);
       fprintf(memf, "\n%s", msg);
    } else {
       fprintf(memf, "%s", msg);
    }
    if (loc2.block) {
       fprintf(memf, " in BB%d:\n", loc2.block->index);
-      aco_print_instr(loc2.instr, memf);
+      aco_print_instr(program->gfx_level, loc2.instr, memf);
    }
    fprintf(memf, "\n\n");
    u_memstream_close(&mem);
diff --git a/src/amd/compiler/tests/test_regalloc.cpp b/src/amd/compiler/tests/test_regalloc.cpp
index 78f5db0564d..7d9bf38413b 100644
--- a/src/amd/compiler/tests/test_regalloc.cpp
+++ b/src/amd/compiler/tests/test_regalloc.cpp
@@ -272,7 +272,7 @@ BEGIN_TEST(regalloc.linear_vgpr.live_range_split.get_reg_impl)
 
    //>> lv1: %5:v[2] = p_parallelcopy %3:v[1] scc:1 scratch:s1
    Pseudo_instruction& parallelcopy = program->blocks[0].instructions[3]->pseudo();
-   aco_print_instr(&parallelcopy, output);
+   aco_print_instr(program->gfx_level, &parallelcopy, output);
    fprintf(output, " scc:%u scratch:s%u\n", parallelcopy.tmp_in_scc, parallelcopy.scratch_sgpr.reg());
 END_TEST