diff --git a/projects/roctracer/inc/roctracer.h b/projects/roctracer/inc/roctracer.h
index 17cda4446d..04cd586d88 100644
--- a/projects/roctracer/inc/roctracer.h
+++ b/projects/roctracer/inc/roctracer.h
@@ -209,6 +209,9 @@ roctracer_status_t roctracer_disable_activity();
 roctracer_status_t roctracer_flush_activity(
     roctracer_pool_t* pool = NULL);                       // memory pool, NULL is a default one
 
+// Mark API
+void roctracer_mark(const char* str);
+
 // Load/Un;oad methods
 // Set properties
 roctracer_status_t roctracer_set_properties(
diff --git a/projects/roctracer/src/core/loader.h b/projects/roctracer/src/core/loader.h
index 46e8a4194b..ad25c99d61 100644
--- a/projects/roctracer/src/core/loader.h
+++ b/projects/roctracer/src/core/loader.h
@@ -8,6 +8,8 @@
 #define LOADER_INSTANTIATE() \
   std::atomic<roctracer::HipLoader*> roctracer::HipLoader::instance_{}; \
   std::atomic<roctracer::HccLoader*> roctracer::HccLoader::instance_{}; \
+  std::atomic<roctracer::KfdLoader*> roctracer::KfdLoader::instance_{}; \
+  std::atomic<roctracer::RocTxLoader*> roctracer::RocTxLoader::instance_{}; \
   roctracer::Loader::mutex_t roctracer::Loader::mutex_;
 
 namespace roctracer {
@@ -139,7 +141,7 @@ class KfdLoader : protected Loader {
     return *instance_;
   }
 
-  KfdLoader() : Loader("libkfd_wrapper.so") {
+  KfdLoader() : Loader("libkfdwrapper64.so") {
     RegisterApiCallback = GetFun<RegisterApiCallback_t>("RegisterApiCallback");
     RemoveApiCallback = GetFun<RemoveApiCallback_t>("RemoveApiCallback");
   }
@@ -151,6 +153,36 @@ class KfdLoader : protected Loader {
   static std::atomic<KfdLoader*> instance_;
 };
 
+// KFD runtime library loader class
+class RocTxLoader : protected Loader {
+  public:
+  typedef bool (RegisterApiCallback_t)(uint32_t op, void* callback, void* arg);
+  typedef bool (RemoveApiCallback_t)(uint32_t op);
+
+  static RocTxLoader& Instance() {
+    RocTxLoader* obj = instance_.load(std::memory_order_acquire);
+    if (obj == NULL) {
+      std::lock_guard<mutex_t> lck(mutex_);
+      if (instance_.load(std::memory_order_relaxed) == NULL) {
+        obj = new RocTxLoader();
+        instance_.store(obj, std::memory_order_release);
+      }
+    }
+    return *instance_;
+  }
+
+  RocTxLoader() : Loader("libroctx64.so") {
+    RegisterApiCallback = GetFun<RegisterApiCallback_t>("RegisterApiCallback");
+    RemoveApiCallback = GetFun<RemoveApiCallback_t>("RemoveApiCallback");
+  }
+
+  RegisterApiCallback_t* RegisterApiCallback;
+  RemoveApiCallback_t* RemoveApiCallback;
+
+  private:
+  static std::atomic<RocTxLoader*> instance_;
+};
+
 } // namespace roctracer
 
 #endif // SRC_CORE_LOADER_H_
diff --git a/projects/roctracer/src/core/roctracer.cpp b/projects/roctracer/src/core/roctracer.cpp
index 4e9c90d410..819aa018f1 100644
--- a/projects/roctracer/src/core/roctracer.cpp
+++ b/projects/roctracer/src/core/roctracer.cpp
@@ -88,6 +88,12 @@ THE SOFTWARE.
   (void)err;                                                                                       \
   return X;
 
+///////////////////////////////////////////////////////////////////////////////////////////////////
+// Mark callback
+//
+typedef void (mark_api_callback_t)(uint32_t domain, uint32_t cid, const void* callback_data, void* arg);
+mark_api_callback_t* mark_api_callback_ptr = NULL;
+
 ///////////////////////////////////////////////////////////////////////////////////////////////////
 // Internal library methods
 //
@@ -923,6 +929,11 @@ PUBLIC_API roctracer_status_t roctracer_flush_activity(roctracer_pool_t* pool) {
   API_METHOD_SUFFIX
 }
 
+// Mark API
+PUBLIC_API void roctracer_mark(const char* str) {
+  if (mark_api_callback_ptr) mark_api_callback_ptr(ACTIVITY_DOMAIN_NUMBER, 0, str, NULL);
+}
+
 // Set properties
 PUBLIC_API roctracer_status_t roctracer_set_properties(
     roctracer_domain_t domain,
@@ -964,7 +975,7 @@ PUBLIC_API roctracer_status_t roctracer_set_properties(
     }
     case ACTIVITY_DOMAIN_HCC_OPS:
     case ACTIVITY_DOMAIN_HIP_API:
-      EXC_RAISING(ROCTRACER_STATUS_BAD_DOMAIN, "properties are not supported, domain ID(" << domain << ")");
+      mark_api_callback_ptr = reinterpret_cast<mark_api_callback_t*>(properties);
     default:
       EXC_RAISING(ROCTRACER_STATUS_BAD_DOMAIN, "invalid domain ID(" << domain << ")");
   }
diff --git a/projects/roctracer/test/run.sh b/projects/roctracer/test/run.sh
index 7d75f910c6..807a0202bc 100755
--- a/projects/roctracer/test/run.sh
+++ b/projects/roctracer/test/run.sh
@@ -26,21 +26,35 @@
 export HSA_TOOLS_REPORT_LOAD_FAILURE=1
 # paths to ROC profiler and oher libraries
 export LD_LIBRARY_PATH=$PWD
+
+# test filter input
+test_filter=-1
+if [ -n "$1" ] ; then
+  test_filter=$1
+fi
+
 # test check routin
 test_status=0
+test_runnum=0
 test_number=0
+xeval_test() {
+  test_number=$test_number
+}
 eval_test() {
   label=$1
   cmdline=$2
-  echo "$label: \"$cmdline\""
-  eval "$cmdline"
-  if [ $? != 0 ] ; then
-    echo "$label: FAILED"
-    test_status=$(($test_status + 1))
-  else
-    echo "$label: PASSED"
+  if [ $test_filter = -1  -o $test_filter = $test_number ] ; then
+    echo "$label: \"$cmdline\""
+    test_runnum=$((test_runnum + 1))
+    eval "$cmdline"
+    if [ $? != 0 ] ; then
+      echo "$label: FAILED"
+      test_status=$(($test_status + 1))
+    else
+      echo "$label: PASSED"
+    fi
   fi
-  test_number=$(($test_number + 1))
+  test_number=$((test_number + 1))
 }
 
 # Standalone test
@@ -80,5 +94,5 @@ eval_test "tool HSA test input" ./test/hsa/ctrl
 #valgrind --tool=massif $tbin
 #ms_print massif.out.<N>
 
-echo "$test_number tests total / $test_status tests failed"
+echo "$test_number tests total / $test_runnum tests run / $test_status tests failed"
 exit $test_status
diff --git a/projects/roctracer/test/tool/tracer_tool.cpp b/projects/roctracer/test/tool/tracer_tool.cpp
index 2bfb3ba915..67ad641a47 100644
--- a/projects/roctracer/test/tool/tracer_tool.cpp
+++ b/projects/roctracer/test/tool/tracer_tool.cpp
@@ -148,6 +148,7 @@ void hsa_activity_callback(
 struct hip_api_trace_entry_t {
   uint32_t valid;
   uint32_t type;
+  uint32_t domain;
   uint32_t cid;
   timestamp_t begin;
   timestamp_t end;
@@ -179,6 +180,7 @@ void hip_api_callback(
     entry->valid = roctracer::TRACE_ENTRY_COMPL;
     entry->type = 0;
     entry->cid = cid;
+    entry->domain = domain;
     entry->begin = hip_begin_timestamp;
     entry->end = end_timestamp;
     entry->pid = GetPid();
@@ -202,46 +204,76 @@ void hip_api_callback(
   }
 }
 
+void mark_api_callback(
+    uint32_t domain,
+    uint32_t cid,
+    const void* callback_data,
+    void* arg)
+{
+  (void)arg;
+  const char* name = reinterpret_cast<const char*>(callback_data);
+
+  const timestamp_t timestamp = timer->timestamp_fn_ns();
+  hip_api_trace_entry_t* entry = hip_api_trace_buffer.GetEntry();
+  entry->valid = roctracer::TRACE_ENTRY_COMPL;
+  entry->type = 0;
+  entry->cid = 0;
+  entry->domain = domain;
+  entry->begin = timestamp;
+  entry->end = timestamp + 1;
+  entry->pid = GetPid();
+  entry->tid = GetTid();
+  entry->data = {};
+  entry->name = name;
+  entry->ptr = NULL;
+}
+
 void hip_api_flush_cb(hip_api_trace_entry_t* entry) {
+  const uint32_t domain = entry->domain;
   const uint32_t cid = entry->cid;
   const hip_api_data_t* data = &(entry->data);
   const timestamp_t begin_timestamp = entry->begin;
   const timestamp_t end_timestamp = entry->end;
   std::ostringstream oss;                                                                        \
 
+  const char* str = (domain < ACTIVITY_DOMAIN_NUMBER) ? roctracer_op_string(domain, cid, 0) : strdup("MARK");
   oss << std::dec <<
-    begin_timestamp << ":" << end_timestamp << " " << entry->pid << ":" << entry->tid << " " << roctracer_op_string(ACTIVITY_DOMAIN_HIP_API, cid, 0);
+    begin_timestamp << ":" << end_timestamp << " " << entry->pid << ":" << entry->tid << " " << str;
 
-  switch (cid) {
-    case HIP_API_ID_hipMemcpy:
-      fprintf(hip_api_file_handle, "%s(dst(%p) src(%p) size(0x%x) kind(%u))\n",
-        oss.str().c_str(),
-        data->args.hipMemcpy.dst,
-        data->args.hipMemcpy.src,
-        (uint32_t)(data->args.hipMemcpy.sizeBytes),
-        (uint32_t)(data->args.hipMemcpy.kind));
-      break;
-    case HIP_API_ID_hipMalloc:
-      fprintf(hip_api_file_handle, "%s(ptr(%p) size(0x%x))\n",
-        oss.str().c_str(),
-        entry->ptr,
-        (uint32_t)(data->args.hipMalloc.size));
-      break;
-    case HIP_API_ID_hipFree:
-      fprintf(hip_api_file_handle, "%s(ptr(%p))\n",
-        oss.str().c_str(),
-        data->args.hipFree.ptr);
-      break;
-    case HIP_API_ID_hipModuleLaunchKernel:
-    case HIP_API_ID_hipExtModuleLaunchKernel:
-    case HIP_API_ID_hipHccModuleLaunchKernel:
-      fprintf(hip_api_file_handle, "%s(kernel(%s) stream(%p))\n",
-        oss.str().c_str(),
-        cxx_demangle(entry->name),
-        data->args.hipModuleLaunchKernel.stream);
-      break;
-    default:
-      fprintf(hip_api_file_handle, "%s()\n", oss.str().c_str());
+  if (domain == ACTIVITY_DOMAIN_HIP_API) {
+    switch (cid) {
+      case HIP_API_ID_hipMemcpy:
+        fprintf(hip_api_file_handle, "%s(dst(%p) src(%p) size(0x%x) kind(%u))\n",
+          oss.str().c_str(),
+          data->args.hipMemcpy.dst,
+          data->args.hipMemcpy.src,
+          (uint32_t)(data->args.hipMemcpy.sizeBytes),
+          (uint32_t)(data->args.hipMemcpy.kind));
+        break;
+      case HIP_API_ID_hipMalloc:
+        fprintf(hip_api_file_handle, "%s(ptr(%p) size(0x%x))\n",
+          oss.str().c_str(),
+          entry->ptr,
+          (uint32_t)(data->args.hipMalloc.size));
+        break;
+      case HIP_API_ID_hipFree:
+        fprintf(hip_api_file_handle, "%s(ptr(%p))\n",
+          oss.str().c_str(),
+          data->args.hipFree.ptr);
+        break;
+      case HIP_API_ID_hipModuleLaunchKernel:
+      case HIP_API_ID_hipExtModuleLaunchKernel:
+      case HIP_API_ID_hipHccModuleLaunchKernel:
+        fprintf(hip_api_file_handle, "%s(kernel(%s) stream(%p))\n",
+          oss.str().c_str(),
+          cxx_demangle(entry->name),
+          data->args.hipModuleLaunchKernel.stream);
+        break;
+      default:
+        fprintf(hip_api_file_handle, "%s()\n", oss.str().c_str());
+    }
+  } else {
+    fprintf(hip_api_file_handle, "%s(\"%s\")\n", oss.str().c_str(), entry->name);
   }
 
   fflush(hip_api_file_handle);
@@ -461,6 +493,8 @@ extern "C" PUBLIC_API bool OnLoad(HsaApiTable* table, uint64_t runtime_version,
     ROCTRACER_CALL(roctracer_enable_domain_activity(ACTIVITY_DOMAIN_HCC_OPS));
     ROCTRACER_CALL(roctracer_enable_domain_activity(ACTIVITY_DOMAIN_HIP_API));
     ROCTRACER_CALL(roctracer_enable_domain_callback(ACTIVITY_DOMAIN_HIP_API, hip_api_callback, NULL));
+
+    roctracer_set_properties(ACTIVITY_DOMAIN_HIP_API, (void*)mark_api_callback);
   }
 
   return roctracer_load(table, runtime_version, failed_tool_count, failed_tool_names);