[buffer_cache, memory, pipeline] Added translation entry structure and update buffer cache handling

src/core/device_memory_manager.h

@@ -127,6 +127,11 @@ public:
     void UpdatePagesCachedBatch(std::span<const std::pair<DAddr, size_t>> ranges, s32 delta);
 
 private:
+    struct TranslationEntry {
+        DAddr guest_page{};
+        u8* host_ptr{};
+    };
+
     // Internal helper that performs the update assuming the caller already holds the necessary lock.
     void UpdatePagesCachedCountNoLock(DAddr addr, size_t size, s32 delta);
 
@@ -195,6 +200,11 @@ private:
     }
 
     Common::VirtualBuffer<VAddr> cpu_backing_address;
+    inline static thread_local TranslationEntry t_slot0{};
+    inline static thread_local TranslationEntry t_slot1{};
+    inline static thread_local TranslationEntry t_slot2{};
+    inline static thread_local TranslationEntry t_slot3{};
+    inline static thread_local u32 cache_cursor = 0;
     using CounterType = u8;
     using CounterAtomicType = std::atomic_uint8_t;
     static constexpr size_t subentries = 8 / sizeof(CounterType);

src/core/device_memory_manager.inc

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
@@ -247,6 +247,10 @@ void DeviceMemoryManager<Traits>::Map(DAddr address, VAddr virtual_address, size
         }
         impl->multi_dev_address.Register(new_dev, start_id);
     }
+    t_slot0 = {};
+    t_slot1 = {};
+    t_slot2 = {};
+    t_slot3 = {};
     if (track) {
         TrackContinuityImpl(address, virtual_address, size, asid);
     }
@@ -278,6 +282,10 @@ void DeviceMemoryManager<Traits>::Unmap(DAddr address, size_t size) {
             compressed_device_addr[phys_addr - 1] = new_start | MULTI_FLAG;
         }
     }
+    t_slot0 = {};
+    t_slot1 = {};
+    t_slot2 = {};
+    t_slot3 = {};
 }
 template <typename Traits>
 void DeviceMemoryManager<Traits>::TrackContinuityImpl(DAddr address, VAddr virtual_address,
@@ -417,6 +425,50 @@ void DeviceMemoryManager<Traits>::WalkBlock(DAddr addr, std::size_t size, auto o
 template <typename Traits>
 void DeviceMemoryManager<Traits>::ReadBlock(DAddr address, void* dest_pointer, size_t size) {
     device_inter->FlushRegion(address, size);
+    const std::size_t page_offset = address & Memory::YUZU_PAGEMASK;
+    if (size <= Memory::YUZU_PAGESIZE - page_offset) {
+        const DAddr guest_page = address & ~static_cast<DAddr>(Memory::YUZU_PAGEMASK);
+        if (t_slot0.guest_page == guest_page && t_slot0.host_ptr != nullptr) {
+            std::memcpy(dest_pointer, t_slot0.host_ptr + page_offset, size);
+            return;
+        }
+        if (t_slot1.guest_page == guest_page && t_slot1.host_ptr != nullptr) {
+            std::memcpy(dest_pointer, t_slot1.host_ptr + page_offset, size);
+            return;
+        }
+        if (t_slot2.guest_page == guest_page && t_slot2.host_ptr != nullptr) {
+            std::memcpy(dest_pointer, t_slot2.host_ptr + page_offset, size);
+            return;
+        }
+        if (t_slot3.guest_page == guest_page && t_slot3.host_ptr != nullptr) {
+            std::memcpy(dest_pointer, t_slot3.host_ptr + page_offset, size);
+            return;
+        }
+
+        const std::size_t page_index = address >> Memory::YUZU_PAGEBITS;
+        const auto phys_addr = compressed_physical_ptr[page_index];
+        if (phys_addr != 0) {
+            auto* const mem_ptr = GetPointerFromRaw<u8>(
+                (static_cast<PAddr>(phys_addr - 1) << Memory::YUZU_PAGEBITS));
+            switch (cache_cursor & 3U) {
+            case 0:
+                t_slot0 = TranslationEntry{.guest_page = guest_page, .host_ptr = mem_ptr};
+                break;
+            case 1:
+                t_slot1 = TranslationEntry{.guest_page = guest_page, .host_ptr = mem_ptr};
+                break;
+            case 2:
+                t_slot2 = TranslationEntry{.guest_page = guest_page, .host_ptr = mem_ptr};
+                break;
+            default:
+                t_slot3 = TranslationEntry{.guest_page = guest_page, .host_ptr = mem_ptr};
+                break;
+            }
+            cache_cursor = (cache_cursor + 1) & 3U;
+            std::memcpy(dest_pointer, mem_ptr + page_offset, size);
+            return;
+        }
+    }
     WalkBlock(
         address, size,
         [&](size_t copy_amount, DAddr current_vaddr) {
@@ -455,6 +507,50 @@ void DeviceMemoryManager<Traits>::WriteBlock(DAddr address, const void* src_poin
 
 template <typename Traits>
 void DeviceMemoryManager<Traits>::ReadBlockUnsafe(DAddr address, void* dest_pointer, size_t size) {
+    const std::size_t page_offset = address & Memory::YUZU_PAGEMASK;
+    if (size <= Memory::YUZU_PAGESIZE - page_offset) {
+        const DAddr guest_page = address & ~static_cast<DAddr>(Memory::YUZU_PAGEMASK);
+        if (t_slot0.guest_page == guest_page && t_slot0.host_ptr != nullptr) {
+            std::memcpy(dest_pointer, t_slot0.host_ptr + page_offset, size);
+            return;
+        }
+        if (t_slot1.guest_page == guest_page && t_slot1.host_ptr != nullptr) {
+            std::memcpy(dest_pointer, t_slot1.host_ptr + page_offset, size);
+            return;
+        }
+        if (t_slot2.guest_page == guest_page && t_slot2.host_ptr != nullptr) {
+            std::memcpy(dest_pointer, t_slot2.host_ptr + page_offset, size);
+            return;
+        }
+        if (t_slot3.guest_page == guest_page && t_slot3.host_ptr != nullptr) {
+            std::memcpy(dest_pointer, t_slot3.host_ptr + page_offset, size);
+            return;
+        }
+
+        const std::size_t page_index = address >> Memory::YUZU_PAGEBITS;
+        const auto phys_addr = compressed_physical_ptr[page_index];
+        if (phys_addr != 0) {
+            auto* const mem_ptr = GetPointerFromRaw<u8>(
+                (static_cast<PAddr>(phys_addr - 1) << Memory::YUZU_PAGEBITS));
+            switch (cache_cursor & 3U) {
+            case 0:
+                t_slot0 = TranslationEntry{.guest_page = guest_page, .host_ptr = mem_ptr};
+                break;
+            case 1:
+                t_slot1 = TranslationEntry{.guest_page = guest_page, .host_ptr = mem_ptr};
+                break;
+            case 2:
+                t_slot2 = TranslationEntry{.guest_page = guest_page, .host_ptr = mem_ptr};
+                break;
+            default:
+                t_slot3 = TranslationEntry{.guest_page = guest_page, .host_ptr = mem_ptr};
+                break;
+            }
+            cache_cursor = (cache_cursor + 1) & 3U;
+            std::memcpy(dest_pointer, mem_ptr + page_offset, size);
+            return;
+        }
+    }
     WalkBlock(
         address, size,
         [&](size_t copy_amount, DAddr current_vaddr) {

src/video_core/buffer_cache/buffer_base.h

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
@@ -39,7 +42,8 @@ public:
     static constexpr u64 BASE_PAGE_SIZE = 1ULL << BASE_PAGE_BITS;
 
     explicit BufferBase(VAddr cpu_addr_, u64 size_bytes_)
-        : cpu_addr{cpu_addr_}, size_bytes{size_bytes_} {}
+        : cpu_addr{cpu_addr_}, cpu_addr_cached{static_cast<DAddr>(cpu_addr_)},
+          size_bytes{size_bytes_} {}
 
     explicit BufferBase(NullBufferParams) {}
 
@@ -97,6 +101,8 @@ public:
         return cpu_addr;
     }
 
+    DAddr cpu_addr_cached = 0;
+
     /// Returns the offset relative to the given CPU address
     /// @pre IsInBounds returns true
     [[nodiscard]] u32 Offset(VAddr other_cpu_addr) const noexcept {

src/video_core/buffer_cache/buffer_cache.h

@@ -382,6 +382,10 @@ void BufferCache<P>::BindHostComputeBuffers() {
     BindHostComputeUniformBuffers();
     BindHostComputeStorageBuffers();
     BindHostComputeTextureBuffers();
+    if (any_buffer_uploaded) {
+        runtime.PostCopyBarrier();
+        any_buffer_uploaded = false;
+    }
 }
 
 template <class P>
@@ -766,45 +770,231 @@ void BufferCache<P>::BindHostIndexBuffer() {
     }
 }
 
+template <class P>
+void BufferCache<P>::BindHostVertexBuffer(u32 index, Buffer& buffer, u32 offset, u32 size,
+                                          u32 stride) {
+    if constexpr (IS_OPENGL) {
+        runtime.BindVertexBuffer(index, buffer, offset, size, stride);
+    } else {
+        runtime.BindVertexBuffer(index, buffer.Handle(), offset, size, stride);
+    }
+}
+
+template <class P>
+Binding& BufferCache<P>::VertexBufferSlot(u32 index) {
+    ASSERT(index < NUM_VERTEX_BUFFERS);
+    switch (index) {
+    case 0:
+        return v_buffer0;
+    case 1:
+        return v_buffer1;
+    case 2:
+        return v_buffer2;
+    case 3:
+        return v_buffer3;
+    case 4:
+        return v_buffer4;
+    case 5:
+        return v_buffer5;
+    case 6:
+        return v_buffer6;
+    case 7:
+        return v_buffer7;
+    case 8:
+        return v_buffer8;
+    case 9:
+        return v_buffer9;
+    case 10:
+        return v_buffer10;
+    case 11:
+        return v_buffer11;
+    case 12:
+        return v_buffer12;
+    case 13:
+        return v_buffer13;
+    case 14:
+        return v_buffer14;
+    case 15:
+        return v_buffer15;
+#ifndef __APPLE__
+    case 16:
+        return v_buffer16;
+    case 17:
+        return v_buffer17;
+    case 18:
+        return v_buffer18;
+    case 19:
+        return v_buffer19;
+    case 20:
+        return v_buffer20;
+    case 21:
+        return v_buffer21;
+    case 22:
+        return v_buffer22;
+    case 23:
+        return v_buffer23;
+    case 24:
+        return v_buffer24;
+    case 25:
+        return v_buffer25;
+    case 26:
+        return v_buffer26;
+    case 27:
+        return v_buffer27;
+    case 28:
+        return v_buffer28;
+    case 29:
+        return v_buffer29;
+    case 30:
+        return v_buffer30;
+    case 31:
+        return v_buffer31;
+#endif
+    default:
+#ifdef __APPLE__
+        return v_buffer15;
+#else
+        return v_buffer31;
+#endif
+    }
+}
+
+template <class P>
+const Binding& BufferCache<P>::VertexBufferSlot(u32 index) const {
+    ASSERT(index < NUM_VERTEX_BUFFERS);
+    switch (index) {
+    case 0:
+        return v_buffer0;
+    case 1:
+        return v_buffer1;
+    case 2:
+        return v_buffer2;
+    case 3:
+        return v_buffer3;
+    case 4:
+        return v_buffer4;
+    case 5:
+        return v_buffer5;
+    case 6:
+        return v_buffer6;
+    case 7:
+        return v_buffer7;
+    case 8:
+        return v_buffer8;
+    case 9:
+        return v_buffer9;
+    case 10:
+        return v_buffer10;
+    case 11:
+        return v_buffer11;
+    case 12:
+        return v_buffer12;
+    case 13:
+        return v_buffer13;
+    case 14:
+        return v_buffer14;
+    case 15:
+        return v_buffer15;
+#ifndef __APPLE__
+    case 16:
+        return v_buffer16;
+    case 17:
+        return v_buffer17;
+    case 18:
+        return v_buffer18;
+    case 19:
+        return v_buffer19;
+    case 20:
+        return v_buffer20;
+    case 21:
+        return v_buffer21;
+    case 22:
+        return v_buffer22;
+    case 23:
+        return v_buffer23;
+    case 24:
+        return v_buffer24;
+    case 25:
+        return v_buffer25;
+    case 26:
+        return v_buffer26;
+    case 27:
+        return v_buffer27;
+    case 28:
+        return v_buffer28;
+    case 29:
+        return v_buffer29;
+    case 30:
+        return v_buffer30;
+    case 31:
+        return v_buffer31;
+#endif
+    default:
+#ifdef __APPLE__
+        return v_buffer15;
+#else
+        return v_buffer31;
+#endif
+    }
+}
+
+template <class P>
+void BufferCache<P>::UpdateVertexBufferSlot(u32 index, const Binding& binding) {
+    Binding& slot = VertexBufferSlot(index);
+    if (slot.device_addr != binding.device_addr || slot.size != binding.size) {
+        ++vertex_buffers_serial;
+    }
+    slot = binding;
+    if (binding.buffer_id != NULL_BUFFER_ID && binding.size != 0) {
+        enabled_vertex_buffers_mask |= (1u << index);
+    } else {
+        enabled_vertex_buffers_mask &= ~(1u << index);
+    }
+}
+
 template <class P>
 void BufferCache<P>::BindHostVertexBuffers() {
-    HostBindings<typename P::Buffer> host_bindings;
-    bool any_valid{false};
     auto& flags = maxwell3d->dirty.flags;
-    for (u32 index = 0; index < NUM_VERTEX_BUFFERS; ++index) {
-        const Binding& binding = channel_state->vertex_buffers[index];
+    u32 enabled_mask = enabled_vertex_buffers_mask;
+    HostBindings<Buffer> bindings{};
+    u32 last_index = std::numeric_limits<u32>::max();
+    const auto flush_bindings = [&]() {
+        if (bindings.buffers.empty()) {
+            return;
+        }
+        bindings.max_index = bindings.min_index + static_cast<u32>(bindings.buffers.size());
+        runtime.BindVertexBuffers(bindings);
+        bindings = HostBindings<Buffer>{};
+        last_index = std::numeric_limits<u32>::max();
+    };
+    while (enabled_mask != 0) {
+        const u32 index = std::countr_zero(enabled_mask);
+        enabled_mask &= (enabled_mask - 1);
+        const Binding& binding = VertexBufferSlot(index);
         Buffer& buffer = slot_buffers[binding.buffer_id];
         TouchBuffer(buffer, binding.buffer_id);
         SynchronizeBuffer(buffer, binding.device_addr, binding.size);
         if (!flags[Dirty::VertexBuffer0 + index]) {
+            flush_bindings();
             continue;
         }
         flags[Dirty::VertexBuffer0 + index] = false;
-
-        host_bindings.min_index = (std::min)(host_bindings.min_index, index);
-        host_bindings.max_index = (std::max)(host_bindings.max_index, index);
-        any_valid = true;
-    }
-
-    if (any_valid) {
-        host_bindings.max_index++;
-        for (u32 index = host_bindings.min_index; index < host_bindings.max_index; index++) {
-            flags[Dirty::VertexBuffer0 + index] = false;
-
-            const Binding& binding = channel_state->vertex_buffers[index];
-            Buffer& buffer = slot_buffers[binding.buffer_id];
-
-            const u32 stride = maxwell3d->regs.vertex_streams[index].stride;
-            const u32 offset = buffer.Offset(binding.device_addr);
-            buffer.MarkUsage(offset, binding.size);
-
-            host_bindings.buffers.push_back(&buffer);
-            host_bindings.offsets.push_back(offset);
-            host_bindings.sizes.push_back(binding.size);
-            host_bindings.strides.push_back(stride);
+        const u32 stride = maxwell3d->regs.vertex_streams[index].stride;
+        const u32 offset = buffer.Offset(binding.device_addr);
+        buffer.MarkUsage(offset, binding.size);
+        if (!bindings.buffers.empty() && index != last_index + 1) {
+            flush_bindings();
         }
-        runtime.BindVertexBuffers(host_bindings);
+        if (bindings.buffers.empty()) {
+            bindings.min_index = index;
+        }
+        bindings.buffers.push_back(&buffer);
+        bindings.offsets.push_back(offset);
+        bindings.sizes.push_back(binding.size);
+        bindings.strides.push_back(stride);
+        last_index = index;
     }
+    flush_bindings();
 }
 
 template <class P>
@@ -1208,17 +1398,20 @@ void BufferCache<P>::UpdateVertexBuffer(u32 index) {
     u32 size = address_size; // TODO: Analyze stride and number of vertices
     if (array.enable == 0 || size == 0 || !device_addr) {
         channel_state->vertex_buffers[index] = NULL_BINDING;
+        UpdateVertexBufferSlot(index, NULL_BINDING);
         return;
     }
     if (!gpu_memory->IsWithinGPUAddressRange(gpu_addr_end) || size >= 64_MiB) {
         size = static_cast<u32>(gpu_memory->MaxContinuousRange(gpu_addr_begin, size));
     }
     const BufferId buffer_id = FindBuffer(*device_addr, size);
-    channel_state->vertex_buffers[index] = Binding{
+    const Binding binding{
         .device_addr = *device_addr,
         .size = size,
         .buffer_id = buffer_id,
     };
+    channel_state->vertex_buffers[index] = binding;
+    UpdateVertexBufferSlot(index, binding);
 }
 
 template <class P>
@@ -1531,12 +1724,12 @@ void BufferCache<P>::TouchBuffer(Buffer& buffer, BufferId buffer_id) noexcept {
 
 template <class P>
 bool BufferCache<P>::SynchronizeBuffer(Buffer& buffer, DAddr device_addr, u32 size) {
-    boost::container::small_vector<BufferCopy, 4> copies;
+    upload_copies.clear();
     u64 total_size_bytes = 0;
     u64 largest_copy = 0;
-    DAddr buffer_start = buffer.CpuAddr();
+    const DAddr buffer_start = buffer.cpu_addr_cached;
     memory_tracker.ForEachUploadRange(device_addr, size, [&](u64 device_addr_out, u64 range_size) {
-        copies.push_back(BufferCopy{
+        upload_copies.push_back(BufferCopy{
             .src_offset = total_size_bytes,
             .dst_offset = device_addr_out - buffer_start,
             .size = range_size,
@@ -1547,8 +1740,9 @@ bool BufferCache<P>::SynchronizeBuffer(Buffer& buffer, DAddr device_addr, u32 si
     if (total_size_bytes == 0) {
         return true;
     }
-    const std::span<BufferCopy> copies_span(copies.data(), copies.size());
+    const std::span<BufferCopy> copies_span(upload_copies.data(), upload_copies.size());
     UploadMemory(buffer, total_size_bytes, largest_copy, copies_span);
+    any_buffer_uploaded = true;
     return false;
 }
 
@@ -1738,6 +1932,7 @@ void BufferCache<P>::DeleteBuffer(BufferId buffer_id, bool do_not_mark) {
         auto& binding = channel_state->vertex_buffers[index];
         if (binding.buffer_id == buffer_id) {
             binding.buffer_id = BufferId{};
+            UpdateVertexBufferSlot(index, binding);
             dirty_vertex_buffers.push_back(index);
         }
     }

src/video_core/buffer_cache/buffer_cache_base.h

@@ -321,6 +321,7 @@ public:
 
     std::recursive_mutex mutex;
     Runtime& runtime;
+    bool any_buffer_uploaded = false;
 
 private:
     template <typename Func>
@@ -373,6 +374,8 @@ private:
 
     void BindHostTransformFeedbackBuffers();
 
+    void BindHostVertexBuffer(u32 index, Buffer& buffer, u32 offset, u32 size, u32 stride);
+
     void BindHostComputeUniformBuffers();
 
     void BindHostComputeStorageBuffers();
@@ -454,6 +457,12 @@ private:
 
     [[nodiscard]] bool HasFastUniformBufferBound(size_t stage, u32 binding_index) const noexcept;
 
+    [[nodiscard]] Binding& VertexBufferSlot(u32 index);
+
+    [[nodiscard]] const Binding& VertexBufferSlot(u32 index) const;
+
+    void UpdateVertexBufferSlot(u32 index, const Binding& binding);
+
     void ClearDownload(DAddr base_addr, u64 size);
 
     void InlineMemoryImplementation(DAddr dest_address, size_t copy_size,
@@ -473,6 +482,45 @@ private:
 
     u32 last_index_count = 0;
 
+    u32 enabled_vertex_buffers_mask = 0;
+    u64 vertex_buffers_serial = 0;
+    Binding v_buffer0{};
+    Binding v_buffer1{};
+    Binding v_buffer2{};
+    Binding v_buffer3{};
+    Binding v_buffer4{};
+    Binding v_buffer5{};
+    Binding v_buffer6{};
+    Binding v_buffer7{};
+    Binding v_buffer8{};
+    Binding v_buffer9{};
+    Binding v_buffer10{};
+    Binding v_buffer11{};
+    Binding v_buffer12{};
+    Binding v_buffer13{};
+    Binding v_buffer14{};
+    Binding v_buffer15{};
+#ifndef __APPLE__
+    Binding v_buffer16{};
+    Binding v_buffer17{};
+    Binding v_buffer18{};
+    Binding v_buffer19{};
+    Binding v_buffer20{};
+    Binding v_buffer21{};
+    Binding v_buffer22{};
+    Binding v_buffer23{};
+    Binding v_buffer24{};
+    Binding v_buffer25{};
+    Binding v_buffer26{};
+    Binding v_buffer27{};
+    Binding v_buffer28{};
+    Binding v_buffer29{};
+    Binding v_buffer30{};
+    Binding v_buffer31{};
+#endif
+
+    boost::container::small_vector<BufferCopy, 4> upload_copies;
+
     MemoryTracker memory_tracker;
     Common::RangeSet<DAddr> uncommitted_gpu_modified_ranges;
     Common::PageBitsetRangeSet<DAddr, CACHING_PAGEBITS, (1ULL << 34)> gpu_modified_ranges;

src/video_core/renderer_opengl/gl_compute_pipeline.cpp

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
@@ -189,6 +189,10 @@ void ComputePipeline::Configure() {
     buffer_cache.runtime.SetEnableStorageBuffers(use_storage_buffers);
     buffer_cache.runtime.SetImagePointers(textures.data(), images.data());
     buffer_cache.BindHostComputeBuffers();
+    if (buffer_cache.any_buffer_uploaded) {
+        buffer_cache.runtime.PostCopyBarrier();
+        buffer_cache.any_buffer_uploaded = false;
+    }
 
     const VideoCommon::ImageViewInOut* views_it{views.data() + num_texture_buffers +
                                                 num_image_buffers};

src/video_core/renderer_opengl/gl_graphics_pipeline.cpp

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
@@ -558,6 +558,10 @@ bool GraphicsPipeline::ConfigureImpl(bool is_indexed) {
     if (image_binding != 0) {
         glBindImageTextures(0, image_binding, images.data());
     }
+    if (buffer_cache.any_buffer_uploaded) {
+        buffer_cache.runtime.PostCopyBarrier();
+        buffer_cache.any_buffer_uploaded = false;
+    }
 
     return true;
 }

src/video_core/renderer_vulkan/vk_compute_pipeline.cpp

@@ -203,6 +203,10 @@ void ComputePipeline::Configure(Tegra::Engines::KeplerCompute& kepler_compute,
 
     buffer_cache.UpdateComputeBuffers();
     buffer_cache.BindHostComputeBuffers();
+    if (buffer_cache.any_buffer_uploaded) {
+        buffer_cache.runtime.PostCopyBarrier();
+        buffer_cache.any_buffer_uploaded = false;
+    }
 
     RescalingPushConstant rescaling;
     const VideoCommon::SamplerId* samplers_it{samplers.data()};

src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp

@@ -496,6 +496,10 @@ bool GraphicsPipeline::ConfigureImpl(bool is_indexed) {
     if constexpr (Spec::enabled_stages[4]) {
         prepare_stage(4);
     }
+    if (buffer_cache.any_buffer_uploaded) {
+        buffer_cache.runtime.PostCopyBarrier();
+        buffer_cache.any_buffer_uploaded = false;
+    }
     texture_cache.UpdateRenderTargets(false);
     texture_cache.CheckFeedbackLoop(views);
     ConfigureDraw(rescaling, render_area);