texture_cache: don't count deleted images still waiting to be destroyed

The collector deletes images into a ring that destroys them TICKS_TO_DESTROY
frames later, and until then the driver still counts their memory. Reading the
driver's usage every frame, the collector saw no drop after deleting and kept
deleting 40+ images a frame in aggressive mode long after it had freed enough:
once a session reached the critical threshold, frames took over 100 ms for
seconds at a time.

Keep the bytes deleted into each slot of the ring and leave them out of the
reported usage until the ring destroys them.

2 files changed, 15 insertions, 0 deletions

diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index c8aaf8dff5..71f89db367 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -250,6 +250,12 @@ void TextureCache<P>::TickFrame() {
     // If we can obtain the memory info, use it instead of the estimate.
     if (runtime.CanReportMemoryUsage()) {
         total_used_memory = runtime.GetDeviceMemoryUsage();
+        // Images the collector deleted keep their memory for TICKS_TO_DESTROY frames: counting
+        // it kept the collector in aggressive mode, deleting 40+ images a frame for frames after
+        // it had freed enough (title 1: frames over 100 ms for seconds at a time).
+        for (const u64 bytes : sentenced_image_bytes) {
+            total_used_memory -= (std::min)(total_used_memory, bytes);
+        }
     }
     // Every 600 frames: device memory in use against the collector's thresholds, and what the
     // images themselves account for (VRAM pressure makes the collector thrash).
@@ -316,6 +322,8 @@ void TextureCache<P>::TickFrame() {
         RunGarbageCollector();
     }
     sentenced_images.Tick();
+    sentenced_image_index = (sentenced_image_index + 1) % TICKS_TO_DESTROY;
+    sentenced_image_bytes[sentenced_image_index] = 0;
     sentenced_framebuffers.Tick();
     sentenced_image_view.Tick();
     TickAsyncDecode();
@@ -2832,6 +2840,8 @@ void TextureCache<P>::DeleteImage(ImageId image_id, bool immediate_delete) {
     }
     // TickFrame sets the total from the driver, which can be under the sum of these estimates.
     total_used_memory -= (std::min)(total_used_memory, ImageMemoryBytes(slot_images[image_id]));
+    const u64 freed_bytes = (image.HasScaled() ? GetScaledImageSizeBytes(image) : 0) +
+                            ImageMemoryBytes(slot_images[image_id]);
     const GPUVAddr gpu_addr = image.gpu_addr;
     const auto alloc_it = image_allocs_table.find(gpu_addr);
     if (alloc_it == image_allocs_table.end()) {
@@ -2897,6 +2907,7 @@ void TextureCache<P>::DeleteImage(ImageId image_id, bool immediate_delete) {
         LOG_INFO(Render_Vulkan, "Retained image for queued download: id={}", image_id.Value());
     } else if (!immediate_delete) {
         sentenced_images.Push(std::move(slot_images[image_id]));
+        sentenced_image_bytes[sentenced_image_index] += freed_bytes;
     }
     slot_images.erase(image_id);
 
diff --git a/src/video_core/texture_cache/texture_cache_base.h b/src/video_core/texture_cache/texture_cache_base.h
index 94102cd261..d51eefeee8 100644
--- a/src/video_core/texture_cache/texture_cache_base.h
+++ b/src/video_core/texture_cache/texture_cache_base.h
@@ -552,6 +552,10 @@ private:
     static constexpr size_t TICKS_TO_DESTROY = 8;
 #endif
     DelayedDestructionRing<Image, TICKS_TO_DESTROY> sentenced_images;
+    /// Device memory of the images in each slot of sentenced_images (same index): deleted, but
+    /// counted in the driver's usage until the ring destroys them.
+    std::array<u64, TICKS_TO_DESTROY> sentenced_image_bytes{};
+    size_t sentenced_image_index = 0;
     DelayedDestructionRing<ImageView, TICKS_TO_DESTROY> sentenced_image_view;
     DelayedDestructionRing<Framebuffer, TICKS_TO_DESTROY> sentenced_framebuffers;