diff --git a/src/d3d11/d3d11_annotation.cpp b/src/d3d11/d3d11_annotation.cpp
index 97d29a10..7c74b7ad 100644
--- a/src/d3d11/d3d11_annotation.cpp
+++ b/src/d3d11/d3d11_annotation.cpp
@@ -6,13 +6,46 @@
 
 namespace dxvk {
 
+  template <bool Register>
+  static void RegisterUserDefinedAnnotation(IDXVKUserDefinedAnnotation* annotation) {
+    using RegistrationFunctionType = void(__stdcall *)(IDXVKUserDefinedAnnotation*);
+    static const int16_t RegisterOrdinal = 28257;
+    static const int16_t UnregisterOrdinal = 28258;
+
+    HMODULE d3d9Module = ::LoadLibraryA("d3d9.dll");
+    if (!d3d9Module) {
+      Logger::info("Unable to find d3d9, some annotations may be missed.");
+      return;
+    }
+
+    const int16_t ordinal = Register ? RegisterOrdinal : UnregisterOrdinal;
+    auto registrationFunction = reinterpret_cast<RegistrationFunctionType>(::GetProcAddress(d3d9Module,
+      reinterpret_cast<const char*>(static_cast<uintptr_t>(ordinal))));
+
+    if (!registrationFunction) {
+      Logger::info("Unable to find DXVK_RegisterAnnotation, some annotations may be missed.");
+      return;
+    }
+
+    registrationFunction(annotation);
+  }
+
   D3D11UserDefinedAnnotation::D3D11UserDefinedAnnotation(D3D11DeviceContext* ctx)
   : m_container(ctx),
-    m_eventDepth(0) { }
+    m_eventDepth(0) {
+    if (m_container->IsAnnotationEnabled())
+      RegisterUserDefinedAnnotation<true>(this);
+  }
 
+  D3D11UserDefinedAnnotation::D3D11UserDefinedAnnotation(const D3D11UserDefinedAnnotation&)
+  {
+    if (m_container->IsAnnotationEnabled())
+      RegisterUserDefinedAnnotation<true>(this);
+  }
 
   D3D11UserDefinedAnnotation::~D3D11UserDefinedAnnotation() {
-
+    if (m_container->IsAnnotationEnabled())
+      RegisterUserDefinedAnnotation<false>(this);
   }
 
 
@@ -39,6 +72,8 @@ namespace dxvk {
     if (!m_container->IsAnnotationEnabled())
       return -1;
 
+    D3D10DeviceLock lock = m_container->LockContext();
+
     m_container->EmitCs([color = Color, labelName = dxvk::str::fromws(Name)](DxvkContext *ctx) {
       VkDebugUtilsLabelEXT label;
       label.sType = VK_STRUCTURE_TYPE_DEBUG_UTILS_LABEL_EXT;
@@ -73,6 +108,8 @@ namespace dxvk {
     if (!m_container->IsAnnotationEnabled())
       return;
 
+    D3D10DeviceLock lock = m_container->LockContext();
+
     m_container->EmitCs([color = Color, labelName = dxvk::str::fromws(Name)](DxvkContext *ctx) {
       VkDebugUtilsLabelEXT label;
       label.sType = VK_STRUCTURE_TYPE_DEBUG_UTILS_LABEL_EXT;
diff --git a/src/d3d11/d3d11_annotation.h b/src/d3d11/d3d11_annotation.h
index 702abe40..06d80af6 100644
--- a/src/d3d11/d3d11_annotation.h
+++ b/src/d3d11/d3d11_annotation.h
@@ -12,6 +12,7 @@ namespace dxvk {
   public:
 
     D3D11UserDefinedAnnotation(D3D11DeviceContext* ctx);
+    D3D11UserDefinedAnnotation(const D3D11UserDefinedAnnotation&);
     ~D3D11UserDefinedAnnotation();
 
     ULONG STDMETHODCALLTYPE AddRef();
diff --git a/src/d3d11/d3d11_context.cpp b/src/d3d11/d3d11_context.cpp
index b9b91efb..4a1bf67b 100644
--- a/src/d3d11/d3d11_context.cpp
+++ b/src/d3d11/d3d11_context.cpp
@@ -16,10 +16,10 @@ namespace dxvk {
           DxvkCsChunkFlags        CsFlags)
   : D3D11DeviceChild<ID3D11DeviceContext4>(pParent),
     m_contextExt(this),
-    m_annotation(this),
     m_multithread(this, false),
     m_device    (Device),
     m_staging   (Device, StagingBufferSize),
+    m_annotation(this),
     m_csFlags   (CsFlags),
     m_csChunk   (AllocCsChunk()),
     m_cmdData   (nullptr) {
diff --git a/src/d3d11/d3d11_context.h b/src/d3d11/d3d11_context.h
index c63e60e4..3bd60a57 100644
--- a/src/d3d11/d3d11_context.h
+++ b/src/d3d11/d3d11_context.h
@@ -693,13 +693,15 @@ namespace dxvk {
   protected:
     
     D3D11DeviceContextExt       m_contextExt;
-    D3D11UserDefinedAnnotation  m_annotation;
     D3D10Multithread            m_multithread;
     
     Rc<DxvkDevice>              m_device;
     Rc<DxvkDataBuffer>          m_updateBuffer;
 
     DxvkStagingBuffer           m_staging;
+   
+    //has to be declared after m_device, as compiler initialize in order of declaration in the class
+    D3D11UserDefinedAnnotation  m_annotation;
 
     DxvkCsChunkFlags            m_csFlags;
     DxvkCsChunkRef              m_csChunk;
diff --git a/src/d3d9/d3d9_main.cpp b/src/d3d9/d3d9_main.cpp
index b6e959dd..367dac9c 100644
--- a/src/d3d9/d3d9_main.cpp
+++ b/src/d3d9/d3d9_main.cpp
@@ -3,7 +3,7 @@
 #include "d3d9_interface.h"
 #include "d3d9_shader_validator.h"
 
-#include "../d3d9/d3d9_annotation.h"
+#include "d3d9_annotation.h"
 
 class D3DFE_PROCESSVERTICES;
 using PSGPERRORID = UINT;
diff --git a/src/d3d9/meson.build b/src/d3d9/meson.build
index 9750cd4c..2cd8951e 100644
--- a/src/d3d9/meson.build
+++ b/src/d3d9/meson.build
@@ -39,7 +39,7 @@ d3d9_src = [
   'd3d9_swvp_emu.cpp',
   'd3d9_format_helpers.cpp',
   'd3d9_hud.cpp',
-  'd3d9_annotation.cpp',
+  'd3d9_annotation.cpp'
 ]
 
 d3d9_dll = shared_library('d3d9'+dll_ext, d3d9_src, glsl_generator.process(d3d9_shaders), d3d9_res,