Add option to control GPU buffer readback

src/common/settings.h

@@ -573,6 +573,13 @@ struct Values {
         false,
 #endif
         "rescale_hack", Category::RendererHacks};
+    SwitchableSetting<bool> enable_gpu_buffer_readback{linkage,
+                                                       false,
+                                                       "enable_gpu_buffer_readback",
+                                                       Category::RendererAdvanced,
+                                                       Specialization::Default,
+                                                       true,
+                                                       true};
 
     SwitchableSetting<bool> use_asynchronous_shaders{linkage, false, "use_asynchronous_shaders",
                                                      Category::RendererHacks};

src/qt_common/config/shared_translation.cpp

@@ -229,6 +229,8 @@ std::unique_ptr<TranslationMap> InitializeTranslations(QObject* parent) {
     INSERT(Settings, dma_accuracy, tr("DMA Accuracy:"),
            tr("Controls the DMA precision accuracy. Safe precision fixes issues in some games but "
               "may degrade performance."));
+    INSERT(Settings, enable_gpu_buffer_readback, tr("Enable GPU buffer readback"),
+           tr("Preserves GPU-modified buffer data by reading it back before uploads.\nSome games require this to render certain effects properly.\nMay cause issues if the hardware cannot handle the additional workload."));
     INSERT(Settings, use_asynchronous_shaders, tr("Enable asynchronous shader compilation"),
            tr("May reduce shader stutter."));
     INSERT(Settings, fast_gpu_time, tr("Fast GPU Time"),

src/video_core/buffer_cache/buffer_cache.h

@@ -1634,15 +1634,17 @@ bool BufferCache<P>::SynchronizeBuffer(Buffer& buffer, DAddr device_addr, u32 si
     if (total_size_bytes == 0) {
         return true;
     }
-    u64 min_offset = (std::numeric_limits<u64>::max)();
-    u64 max_offset = 0;
-    for (const auto& copy : upload_copies) {
-        min_offset = (std::min)(min_offset, copy.dst_offset);
-        max_offset = (std::max)(max_offset, copy.dst_offset + copy.size);
-    }
-    const DAddr sync_addr = buffer.CpuAddr() + min_offset;
-    const u64 sync_size = max_offset - min_offset;
-    DownloadBufferMemory(buffer, sync_addr, sync_size);
+    if (Settings::values.enable_gpu_buffer_readback.GetValue()) {
+        u64 min_offset = (std::numeric_limits<u64>::max)();
+        u64 max_offset = 0;
+        for (const auto& copy : upload_copies) {
+            min_offset = (std::min)(min_offset, copy.dst_offset);
+            max_offset = (std::max)(max_offset, copy.dst_offset + copy.size);
+        }
+        const DAddr sync_addr = buffer.CpuAddr() + min_offset;
+        const u64 sync_size = max_offset - min_offset;
+        DownloadBufferMemory(buffer, sync_addr, sync_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;