diff --git a/app/src/main/cpp/skyline/gpu/texture/texture.cpp b/app/src/main/cpp/skyline/gpu/texture/texture.cpp
index 6e50ac13..32a6af9a 100644
--- a/app/src/main/cpp/skyline/gpu/texture/texture.cpp
+++ b/app/src/main/cpp/skyline/gpu/texture/texture.cpp
@@ -825,15 +825,16 @@ namespace skyline::gpu {
         WaitOnBacking();
 
         if (tiling == vk::ImageTiling::eOptimal || !std::holds_alternative<memory::Image>(backing)) {
-            auto stagingBuffer{gpu.memory.AllocateStagingBuffer(surfaceSize)};
+            if (!downloadStagingBuffer)
+                downloadStagingBuffer = gpu.memory.AllocateStagingBuffer(surfaceSize);
 
             WaitOnFence();
             auto lCycle{gpu.scheduler.Submit([&](vk::raii::CommandBuffer &commandBuffer) {
-                CopyIntoStagingBuffer(commandBuffer, stagingBuffer);
+                CopyIntoStagingBuffer(commandBuffer, downloadStagingBuffer);
             })};
             lCycle->Wait(); // We block till the copy is complete
 
-            CopyToGuest(stagingBuffer->data());
+            CopyToGuest(downloadStagingBuffer->data());
         } else if (tiling == vk::ImageTiling::eLinear) {
             // We can optimize linear texture sync on a UMA by mapping the texture onto the CPU and copying directly from it rather than using a staging buffer
             WaitOnFence();
diff --git a/app/src/main/cpp/skyline/gpu/texture/texture.h b/app/src/main/cpp/skyline/gpu/texture/texture.h
index 0c0cb6ff..641a1112 100644
--- a/app/src/main/cpp/skyline/gpu/texture/texture.h
+++ b/app/src/main/cpp/skyline/gpu/texture/texture.h
@@ -406,6 +406,8 @@ namespace skyline::gpu {
 
         std::vector<TextureViewStorage> views;
 
+        std::shared_ptr<memory::StagingBuffer> downloadStagingBuffer{};
+
         u32 lastRenderPassIndex{}; //!< The index of the last render pass that used this texture
         texture::RenderPassUsage lastRenderPassUsage{texture::RenderPassUsage::None}; //!< The type of usage in the last render pass