From 7ae49cd06aa6466922c77026d3cbedfb30034b88 Mon Sep 17 00:00:00 2001
From: IndecisiveTurtle <47210458+raphaelthegreat@users.noreply.github.com>
Date: Tue, 15 Jul 2025 22:07:12 +0300
Subject: [PATCH] vk_rasterizer: Use xor as heuristic for HTILE clear

---
 src/shader_recompiler/frontend/instruction.h  |  2 --
 src/shader_recompiler/info.h                  |  1 +
 .../ir/passes/shader_info_collection_pass.cpp |  3 +++
 .../renderer_vulkan/vk_rasterizer.cpp         | 19 ++++++++++++++-----
 src/video_core/texture_cache/texture_cache.h  |  4 ++++
 5 files changed, 22 insertions(+), 7 deletions(-)

diff --git a/src/shader_recompiler/frontend/instruction.h b/src/shader_recompiler/frontend/instruction.h
index 7c2e0bd1e..f4e7bc9f2 100644
--- a/src/shader_recompiler/frontend/instruction.h
+++ b/src/shader_recompiler/frontend/instruction.h
@@ -3,8 +3,6 @@
 
 #pragma once
 
-#include <limits>
-#include "common/bit_field.h"
 #include "shader_recompiler/frontend/opcodes.h"
 
 namespace Shader::Gcn {
diff --git a/src/shader_recompiler/info.h b/src/shader_recompiler/info.h
index bb5c88584..11dd9c05e 100644
--- a/src/shader_recompiler/info.h
+++ b/src/shader_recompiler/info.h
@@ -222,6 +222,7 @@ struct Info {
     VAddr pgm_base;
     bool has_storage_images{};
     bool has_discard{};
+    bool has_bitwise_xor{};
     bool has_image_gather{};
     bool has_image_query{};
     bool uses_buffer_atomic_float_min_max{};
diff --git a/src/shader_recompiler/ir/passes/shader_info_collection_pass.cpp b/src/shader_recompiler/ir/passes/shader_info_collection_pass.cpp
index 079827866..8f0e61da2 100644
--- a/src/shader_recompiler/ir/passes/shader_info_collection_pass.cpp
+++ b/src/shader_recompiler/ir/passes/shader_info_collection_pass.cpp
@@ -95,6 +95,9 @@ void Visit(Info& info, const IR::Inst& inst) {
     case IR::Opcode::DiscardCond:
         info.has_discard = true;
         break;
+    case IR::Opcode::BitwiseXor32:
+        info.has_bitwise_xor = true;
+        break;
     case IR::Opcode::ImageGather:
     case IR::Opcode::ImageGatherDref:
         info.has_image_gather = true;
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
index b6130e873..c5f894b10 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
@@ -503,9 +503,13 @@ bool Rasterizer::IsComputeMetaClear(const Pipeline* pipeline) {
         return false;
     }
 
+    // Most of the time when a metadata is updated with a shader it gets cleared. It means
+    // we can skip the whole dispatch and update the tracked state instead. Also, it is not
+    // intended to be consumed and in such rare cases (e.g. HTile introspection, CRAA) we
+    // will need its full emulation anyways.
     const auto& info = pipeline->GetStage(Shader::LogicalStage::Compute);
 
-    // Assume if a shader reads and writes metas at the same time, it is a copy shader.
+    // Assume if a shader reads metadata, it is a copy shader.
     for (const auto& desc : info.buffers) {
         const VAddr address = desc.GetSharp(info).base_address;
         if (!desc.IsSpecial() && !desc.is_written && texture_cache.IsMeta(address)) {
@@ -513,10 +517,15 @@ bool Rasterizer::IsComputeMetaClear(const Pipeline* pipeline) {
         }
     }
 
-    // Most of the time when a metadata is updated with a shader it gets cleared. It means
-    // we can skip the whole dispatch and update the tracked state instead. Also, it is not
-    // intended to be consumed and in such rare cases (e.g. HTile introspection, CRAA) we
-    // will need its full emulation anyways.
+    // Metadata surfaces are tiled and thus need address calculation to be written properly.
+    // If a shader wants to encode HTILE, for example, from a depth image it will have to compute
+    // proper tile address from dispatch invocation id. This address calculation contains an xor
+    // operation so use it as a heuristic for metadata writes that are probably not clears.
+    if (info.has_bitwise_xor) {
+        return false;
+    }
+
+    // Assume if a shader writes metadata without address calculation, it is a clear shader.
     for (const auto& desc : info.buffers) {
         const VAddr address = desc.GetSharp(info).base_address;
         if (!desc.IsSpecial() && desc.is_written && texture_cache.ClearMeta(address)) {
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index ff8ffb61c..9a9679c0a 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -161,10 +161,12 @@ public:
     /// Registers an image view for provided image
     ImageView& RegisterImageView(ImageId image_id, const ImageViewInfo& view_info);
 
+    /// Returns true if the specified address is a metadata surface.
     bool IsMeta(VAddr address) const {
         return surface_metas.contains(address);
     }
 
+    /// Returns true if a slice of the specified metadata surface has been cleared.
     bool IsMetaCleared(VAddr address, u32 slice) const {
         const auto& it = surface_metas.find(address);
         if (it != surface_metas.end()) {
@@ -173,6 +175,7 @@ public:
         return false;
     }
 
+    /// Clears all slices of the specified metadata surface.
     bool ClearMeta(VAddr address) {
         auto it = surface_metas.find(address);
         if (it != surface_metas.end()) {
@@ -182,6 +185,7 @@ public:
         return false;
     }
 
+    /// Updates the state of a slice of the specified metadata surface.
     bool TouchMeta(VAddr address, u32 slice, bool is_clear) {
         auto it = surface_metas.find(address);
         if (it != surface_metas.end()) {