video_core: cut per-draw work on the GPU back thread in crowded scenes (#374)

* video_core: demote the texture cache's per-draw debug logs to trace

The invalid sampler and image view index messages fire for unused
binding slots on every draw. Filtered LOG_DEBUG messages still cost a
call and argument packing: in a Fire Emblem: Fortune's Weave city scene
(about 2,100 draws per frame) they were 3.1% of the GPUBack thread's
samples. LOG_TRACE compiles them out of release builds.

Assisted-by: Claude Opus 5.5 (profiling, porting and drafting)
Signed-off-by: Duck912 <350528965+Duck912@users.noreply.github.com>

* texture_cache: read a descriptor with one page lookup

DescriptorTable::Read runs for every texture and sampler of every draw,
and it read each 32-byte TIC/TSC entry through the generic
ReadBlockUnsafe page walk. In a Fire Emblem: Fortune's Weave city scene
(about 2,100 draws per frame) that was 5.0% of the GPUBack thread's
samples. An entry within one page is contiguous in host memory, so it
is now a single page lookup and memcpy. Entries that cross a page and
pages without host memory still take the block read, so unmapped and
unbacked entries still read as zero.

The same change in another Eden-based build cut these reads from 6.4%
to 0.4% of its GPU thread.

Assisted-by: Claude Opus 5.5 (profiling, porting and drafting)
Signed-off-by: Duck912 <350528965+Duck912@users.noreply.github.com>

---------

Signed-off-by: Duck912 <350528965+Duck912@users.noreply.github.com>

2 files changed, 13 insertions, 3 deletions

diff --git a/src/video_core/texture_cache/descriptor_table.h b/src/video_core/texture_cache/descriptor_table.h
index a441e572de..2c768b72fc 100644
--- a/src/video_core/texture_cache/descriptor_table.h
+++ b/src/video_core/texture_cache/descriptor_table.h
@@ -7,12 +7,14 @@
 #pragma once
 
 #include <algorithm>
+#include <cstring>
 #include <vector>
 
 #include "common/alignment.h"
 #include "common/common_types.h"
 #include "common/div_ceil.h"
 #include "common/assert.h"
+#include "core/memory.h"
 #include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
 
@@ -44,7 +46,15 @@ public:
         }
         const GPUVAddr gpu_addr = current_gpu_addr + index * sizeof(T);
         std::pair<T, bool> result;
-        gpu_memory.ReadBlockUnsafe(gpu_addr, std::addressof(result.first), sizeof(T));
+        // This runs for every texture and sampler of every draw. An entry within one page is
+        // contiguous in host memory, so a single page lookup replaces the generic block read.
+        const bool one_page =
+            (gpu_addr & Core::Memory::YUZU_PAGEMASK) + sizeof(T) <= Core::Memory::YUZU_PAGESIZE;
+        if (const u8* const host = one_page ? gpu_memory.GetPointer(gpu_addr) : nullptr) {
+            std::memcpy(std::addressof(result.first), host, sizeof(T));
+        } else {
+            gpu_memory.ReadBlockUnsafe(gpu_addr, std::addressof(result.first), sizeof(T));
+        }
         if ((read_descriptors[index / 64] & (1ULL << (index % 64))) != 0) {
             result.second = result.first != descriptors[index];
         } else {
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index abb22b9bfb..ea82ab02e7 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -580,7 +580,7 @@ template <class P>
 SamplerId TextureCache<P>::GetSamplerId(u32 index, bool compute) {
     auto& table = compute ? channel_state->compute_sampler_table : channel_state->graphics_sampler_table;
     if (index > table.current_limit) {
-        LOG_DEBUG(HW_GPU, "Invalid sampler index={}", index);
+        LOG_TRACE(HW_GPU, "Invalid sampler index={}", index);
         return NULL_SAMPLER_ID;
     }
     auto const map_index = index | (compute ? Common::SlotId::TAGGED_VALUE : 0);
@@ -812,7 +812,7 @@ ImageViewId TextureCache<P>::VisitImageView(u32 index, bool compute) {
     index = Tegra::Texture::TextureHeaderIndex(index, table.current_limit,
                                                gpu_memory->UsesNativeTiling());
     if (index > table.current_limit) {
-        LOG_DEBUG(HW_GPU, "Invalid image view index={}", index);
+        LOG_TRACE(HW_GPU, "Invalid image view index={}", index);
         return NULL_IMAGE_VIEW_ID;
     }
     auto const map_index = index | (compute ? Common::SlotId::TAGGED_VALUE : 0);