diff --git a/intern/ghost/intern/GHOST_ContextVK.cc b/intern/ghost/intern/GHOST_ContextVK.cc
index c8e3dccfbae..37db0353177 100644
--- a/intern/ghost/intern/GHOST_ContextVK.cc
+++ b/intern/ghost/intern/GHOST_ContextVK.cc
@@ -603,7 +603,7 @@ GHOST_ContextVK::GHOST_ContextVK(const GHOST_ContextParams &context_params,
       hdr_info_(hdr_info),
       surface_(VK_NULL_HANDLE),
       swapchain_(VK_NULL_HANDLE),
-      frame_data_(GHOST_FRAMES_IN_FLIGHT),
+      frame_data_(2),
       render_frame_(0),
       use_hdr_swapchain_(false)
 {
@@ -1160,7 +1160,10 @@ GHOST_TSuccess GHOST_ContextVK::recreateSwapchain(bool use_hdr_swapchain)
   /* Some platforms require a minimum amount of render frames that is larger than we expect. When
    * that happens we should increase the number of frames in flight. We could also consider
    * splitting the frame in flight and image specific data. */
-  assert(actual_image_count <= GHOST_FRAMES_IN_FLIGHT);
+  if (actual_image_count > frame_data_.size()) {
+    CLOG_TRACE(&LOG, "Vulkan: Increasing frame data to %u frames", actual_image_count);
+    frame_data_.resize(actual_image_count);
+  }
   swapchain_images_.resize(actual_image_count);
   std::vector<VkImage> swapchain_images(actual_image_count);
   vkGetSwapchainImagesKHR(device, swapchain_, &actual_image_count, swapchain_images.data());
diff --git a/intern/ghost/intern/GHOST_ContextVK.hh b/intern/ghost/intern/GHOST_ContextVK.hh
index 1b0c1afc51c..f5fcc30a540 100644
--- a/intern/ghost/intern/GHOST_ContextVK.hh
+++ b/intern/ghost/intern/GHOST_ContextVK.hh
@@ -91,15 +91,6 @@ struct GHOST_Frame {
   void destroy(VkDevice vk_device);
 };
 
-/**
- * The number of frames that GHOST manages.
- *
- * This must be kept in sync with any frame-aligned resources in the
- * Vulkan backend. Notably, VKThreadData::resource_pools_count must
- * match this value.
- */
-constexpr static uint32_t GHOST_FRAMES_IN_FLIGHT = 5;
-
 class GHOST_ContextVK : public GHOST_Context {
   friend class GHOST_XrGraphicsBindingVulkan;
   friend class GHOST_XrGraphicsBindingVulkanD3D;
diff --git a/source/blender/gpu/vulkan/vk_context.cc b/source/blender/gpu/vulkan/vk_context.cc
index adc29516d95..8947f41ef38 100644
--- a/source/blender/gpu/vulkan/vk_context.cc
+++ b/source/blender/gpu/vulkan/vk_context.cc
@@ -56,15 +56,11 @@ VKContext::~VKContext()
   this->process_frame_timings();
 }
 
-void VKContext::sync_backbuffer(bool cycle_resource_pool)
+void VKContext::sync_backbuffer()
 {
   if (ghost_window_) {
     GHOST_VulkanSwapChainData swap_chain_data = {};
     GHOST_GetVulkanSwapChainFormat((GHOST_WindowHandle)ghost_window_, &swap_chain_data);
-    VKThreadData &thread_data = thread_data_.value().get();
-    if (cycle_resource_pool) {
-      thread_data.resource_pool_next();
-    }
 
     const bool reset_framebuffer = swap_chain_format_.format !=
                                        swap_chain_data.surface_format.format ||
@@ -132,7 +128,7 @@ void VKContext::activate()
 
   is_active_ = true;
 
-  sync_backbuffer(false);
+  sync_backbuffer();
 
   immActivate();
 }
@@ -216,7 +212,7 @@ VKDescriptorPools &VKContext::descriptor_pools_get()
 
 VKDescriptorSetTracker &VKContext::descriptor_set_get()
 {
-  return thread_data_.value().get().resource_pool_get().descriptor_set;
+  return thread_data_.value().get().descriptor_set;
 }
 
 VKStateManager &VKContext::state_manager_get() const
@@ -451,7 +447,7 @@ void VKContext::swap_buffers_pre_handler(const GHOST_VulkanSwapChainData &swap_c
 
 void VKContext::swap_buffers_post_handler()
 {
-  sync_backbuffer(true);
+  sync_backbuffer();
 }
 
 void VKContext::specialization_constants_set(
diff --git a/source/blender/gpu/vulkan/vk_context.hh b/source/blender/gpu/vulkan/vk_context.hh
index ed4a61ee1fa..bce610762ce 100644
--- a/source/blender/gpu/vulkan/vk_context.hh
+++ b/source/blender/gpu/vulkan/vk_context.hh
@@ -137,7 +137,7 @@ class VKContext : public Context, NonCopyable {
                             VKVertexAttributeObject &vao,
                             render_graph::VKPipelineData &r_pipeline_data);
 
-  void sync_backbuffer(bool cycle_resource_pool);
+  void sync_backbuffer();
 
   static VKContext *get()
   {
diff --git a/source/blender/gpu/vulkan/vk_device.cc b/source/blender/gpu/vulkan/vk_device.cc
index f8014a04818..4c1f0ebdb12 100644
--- a/source/blender/gpu/vulkan/vk_device.cc
+++ b/source/blender/gpu/vulkan/vk_device.cc
@@ -645,11 +645,6 @@ void VKDevice::debug_print()
     const bool is_main = pthread_equal(thread_data->thread_id, pthread_self());
     os << "ThreadData" << (is_main ? " (main-thread)" : "") << ")\n";
     os << " Rendering_depth: " << thread_data->rendering_depth << "\n";
-    for (int resource_pool_index : IndexRange(thread_data->resource_pools.size())) {
-      const bool is_active = thread_data->resource_pool_index == resource_pool_index;
-      os << " Resource Pool (index=" << resource_pool_index << (is_active ? " active" : "")
-         << ")\n";
-    }
   }
   os << "Discard pool\n";
   debug_print(os, orphaned_data);
diff --git a/source/blender/gpu/vulkan/vk_device.hh b/source/blender/gpu/vulkan/vk_device.hh
index 29272f31fea..1efe816afa2 100644
--- a/source/blender/gpu/vulkan/vk_device.hh
+++ b/source/blender/gpu/vulkan/vk_device.hh
@@ -106,26 +106,12 @@ struct VKWorkarounds {
  * Shared resources between contexts that run in the same thread.
  */
 class VKThreadData : public NonCopyable, NonMovable {
-  /**
-   * The number of resource pools is aligned to the number of frames
-   * in flight used by GHOST. Therefore, this constant *must* always
-   * match GHOST_ContextVK's GHOST_FRAMES_IN_FLIGHT.
-   */
-  static constexpr uint32_t resource_pools_count = 5;
-
  public:
   /** Thread ID this instance belongs to. */
   pthread_t thread_id;
-  /**
-   * Index of the active resource pool. Is in sync with the active swap-chain image or cycled when
-   * rendering.
-   *
-   * NOTE: Initialized to `UINT32_MAX` to detect first change.
-   */
-  uint32_t resource_pool_index = UINT32_MAX;
-  std::array<VKResourcePool, resource_pools_count> resource_pools;
-
   VKDescriptorPools descriptor_pools;
+  VKDescriptorSetTracker descriptor_set;
+
   /**
    * The current rendering depth.
    *
@@ -137,28 +123,6 @@ class VKThreadData : public NonCopyable, NonMovable {
   int32_t rendering_depth = 0;
 
   VKThreadData(VKDevice &device, pthread_t thread_id);
-
-  /**
-   * Get the active resource pool.
-   */
-  VKResourcePool &resource_pool_get()
-  {
-    if (resource_pool_index >= resource_pools.size()) {
-      return resource_pools[0];
-    }
-    return resource_pools[resource_pool_index];
-  }
-
-  /** Activate the next resource pool. */
-  void resource_pool_next()
-  {
-    if (resource_pool_index == UINT32_MAX) {
-      resource_pool_index = 1;
-    }
-    else {
-      resource_pool_index = (resource_pool_index + 1) % resource_pools_count;
-    }
-  }
 };
 
 class VKDevice : public NonCopyable {
diff --git a/source/blender/gpu/vulkan/vk_resource_pool.hh b/source/blender/gpu/vulkan/vk_resource_pool.hh
index 3afc189303d..770433d9e0e 100644
--- a/source/blender/gpu/vulkan/vk_resource_pool.hh
+++ b/source/blender/gpu/vulkan/vk_resource_pool.hh
@@ -143,9 +143,4 @@ class VKDiscardPool {
   static VKDiscardPool &discard_pool_get();
 };
 
-class VKResourcePool {
-
- public:
-  VKDescriptorSetTracker descriptor_set;
-};
 }  // namespace blender::gpu