summaryrefslogtreecommitdiffstats
path: root/src/video_core
diff options
context:
space:
mode:
authorFernando S <fsahmkow27@gmail.com>2021-08-10 12:23:05 +0200
committerGitHub <noreply@github.com>2021-08-10 12:23:05 +0200
commit6a082df427df37ffa6d3e7601549327a0e9b24d1 (patch)
tree4b51ec28b4abb2a2666a2994b8ca35229b6bc241 /src/video_core
parentMerge pull request #6837 from german77/no-pause-screenshot (diff)
parenttexture_cache: Address ameerj's review (diff)
downloadyuzu-6a082df427df37ffa6d3e7601549327a0e9b24d1.tar
yuzu-6a082df427df37ffa6d3e7601549327a0e9b24d1.tar.gz
yuzu-6a082df427df37ffa6d3e7601549327a0e9b24d1.tar.bz2
yuzu-6a082df427df37ffa6d3e7601549327a0e9b24d1.tar.lz
yuzu-6a082df427df37ffa6d3e7601549327a0e9b24d1.tar.xz
yuzu-6a082df427df37ffa6d3e7601549327a0e9b24d1.tar.zst
yuzu-6a082df427df37ffa6d3e7601549327a0e9b24d1.zip
Diffstat (limited to 'src/video_core')
-rw-r--r--src/video_core/CMakeLists.txt3
-rw-r--r--src/video_core/renderer_opengl/gl_graphics_pipeline.cpp2
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.cpp2
-rw-r--r--src/video_core/renderer_opengl/gl_texture_cache.cpp4
-rw-r--r--src/video_core/renderer_opengl/gl_texture_cache.h2
-rw-r--r--src/video_core/renderer_opengl/gl_texture_cache_base.cpp10
-rw-r--r--src/video_core/renderer_vulkan/vk_rasterizer.cpp2
-rw-r--r--src/video_core/renderer_vulkan/vk_texture_cache.cpp2
-rw-r--r--src/video_core/renderer_vulkan/vk_texture_cache.h2
-rw-r--r--src/video_core/renderer_vulkan/vk_texture_cache_base.cpp10
-rw-r--r--src/video_core/texture_cache/image_view_info.cpp4
-rw-r--r--src/video_core/texture_cache/texture_cache.h419
-rw-r--r--src/video_core/texture_cache/texture_cache_base.h385
13 files changed, 420 insertions, 427 deletions
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index 1eb67c051..2f6cdd216 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -97,6 +97,7 @@ add_library(video_core STATIC
renderer_opengl/gl_stream_buffer.h
renderer_opengl/gl_texture_cache.cpp
renderer_opengl/gl_texture_cache.h
+ renderer_opengl/gl_texture_cache_base.cpp
renderer_opengl/gl_query_cache.cpp
renderer_opengl/gl_query_cache.h
renderer_opengl/maxwell_to_gl.h
@@ -155,6 +156,7 @@ add_library(video_core STATIC
renderer_vulkan/vk_swapchain.h
renderer_vulkan/vk_texture_cache.cpp
renderer_vulkan/vk_texture_cache.h
+ renderer_vulkan/vk_texture_cache_base.cpp
renderer_vulkan/vk_update_descriptor.cpp
renderer_vulkan/vk_update_descriptor.h
shader_cache.cpp
@@ -186,6 +188,7 @@ add_library(video_core STATIC
texture_cache/samples_helper.h
texture_cache/slot_vector.h
texture_cache/texture_cache.h
+ texture_cache/texture_cache_base.h
texture_cache/types.h
texture_cache/util.cpp
texture_cache/util.h
diff --git a/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp b/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp
index fac0034fb..bccb37a58 100644
--- a/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp
+++ b/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp
@@ -15,7 +15,7 @@
#include "video_core/renderer_opengl/gl_shader_util.h"
#include "video_core/renderer_opengl/gl_state_tracker.h"
#include "video_core/shader_notify.h"
-#include "video_core/texture_cache/texture_cache.h"
+#include "video_core/texture_cache/texture_cache_base.h"
#if defined(_MSC_VER) && defined(NDEBUG)
#define LAMBDA_FORCEINLINE [[msvc::forceinline]]
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 41d2b73f4..b909c387e 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -32,7 +32,7 @@
#include "video_core/renderer_opengl/maxwell_to_gl.h"
#include "video_core/renderer_opengl/renderer_opengl.h"
#include "video_core/shader_cache.h"
-#include "video_core/texture_cache/texture_cache.h"
+#include "video_core/texture_cache/texture_cache_base.h"
namespace OpenGL {
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp
index c373c9cb4..b0aee6cc1 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp
@@ -18,10 +18,8 @@
#include "video_core/renderer_opengl/maxwell_to_gl.h"
#include "video_core/renderer_opengl/util_shaders.h"
#include "video_core/surface.h"
-#include "video_core/texture_cache/format_lookup_table.h"
+#include "video_core/texture_cache/formatter.h"
#include "video_core/texture_cache/samples_helper.h"
-#include "video_core/texture_cache/texture_cache.h"
-#include "video_core/textures/decoders.h"
namespace OpenGL {
namespace {
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h
index 921072ebe..4a4f6301c 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.h
+++ b/src/video_core/renderer_opengl/gl_texture_cache.h
@@ -12,7 +12,7 @@
#include "shader_recompiler/shader_info.h"
#include "video_core/renderer_opengl/gl_resource_manager.h"
#include "video_core/renderer_opengl/util_shaders.h"
-#include "video_core/texture_cache/texture_cache.h"
+#include "video_core/texture_cache/texture_cache_base.h"
namespace OpenGL {
diff --git a/src/video_core/renderer_opengl/gl_texture_cache_base.cpp b/src/video_core/renderer_opengl/gl_texture_cache_base.cpp
new file mode 100644
index 000000000..385358fea
--- /dev/null
+++ b/src/video_core/renderer_opengl/gl_texture_cache_base.cpp
@@ -0,0 +1,10 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "video_core/renderer_opengl/gl_texture_cache.h"
+#include "video_core/texture_cache/texture_cache.h"
+
+namespace VideoCommon {
+template class VideoCommon::TextureCache<OpenGL::TextureCacheParams>;
+}
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
index 23cef2996..3ac18ea54 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
@@ -32,7 +32,7 @@
#include "video_core/renderer_vulkan/vk_texture_cache.h"
#include "video_core/renderer_vulkan/vk_update_descriptor.h"
#include "video_core/shader_cache.h"
-#include "video_core/texture_cache/texture_cache.h"
+#include "video_core/texture_cache/texture_cache_base.h"
#include "video_core/vulkan_common/vulkan_device.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.cpp b/src/video_core/renderer_vulkan/vk_texture_cache.cpp
index 8e029bcb3..8f4df7122 100644
--- a/src/video_core/renderer_vulkan/vk_texture_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_texture_cache.cpp
@@ -19,6 +19,8 @@
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
#include "video_core/renderer_vulkan/vk_texture_cache.h"
+#include "video_core/texture_cache/formatter.h"
+#include "video_core/texture_cache/samples_helper.h"
#include "video_core/vulkan_common/vulkan_device.h"
#include "video_core/vulkan_common/vulkan_memory_allocator.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.h b/src/video_core/renderer_vulkan/vk_texture_cache.h
index 0b73d55f8..5fe6b7ba3 100644
--- a/src/video_core/renderer_vulkan/vk_texture_cache.h
+++ b/src/video_core/renderer_vulkan/vk_texture_cache.h
@@ -9,7 +9,7 @@
#include "shader_recompiler/shader_info.h"
#include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
-#include "video_core/texture_cache/texture_cache.h"
+#include "video_core/texture_cache/texture_cache_base.h"
#include "video_core/vulkan_common/vulkan_memory_allocator.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
diff --git a/src/video_core/renderer_vulkan/vk_texture_cache_base.cpp b/src/video_core/renderer_vulkan/vk_texture_cache_base.cpp
new file mode 100644
index 000000000..44e688342
--- /dev/null
+++ b/src/video_core/renderer_vulkan/vk_texture_cache_base.cpp
@@ -0,0 +1,10 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "video_core/renderer_vulkan/vk_texture_cache.h"
+#include "video_core/texture_cache/texture_cache.h"
+
+namespace VideoCommon {
+template class VideoCommon::TextureCache<Vulkan::TextureCacheParams>;
+}
diff --git a/src/video_core/texture_cache/image_view_info.cpp b/src/video_core/texture_cache/image_view_info.cpp
index faf5b151f..6527e14c8 100644
--- a/src/video_core/texture_cache/image_view_info.cpp
+++ b/src/video_core/texture_cache/image_view_info.cpp
@@ -6,7 +6,7 @@
#include "common/assert.h"
#include "video_core/texture_cache/image_view_info.h"
-#include "video_core/texture_cache/texture_cache.h"
+#include "video_core/texture_cache/texture_cache_base.h"
#include "video_core/texture_cache/types.h"
#include "video_core/textures/texture.h"
@@ -14,6 +14,8 @@ namespace VideoCommon {
namespace {
+using Tegra::Texture::TextureType;
+
constexpr u8 RENDER_TARGET_SWIZZLE = std::numeric_limits<u8>::max();
[[nodiscard]] u8 CastSwizzle(SwizzleSource source) {
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index f34c9d9ca..a087498ff 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -4,48 +4,11 @@
#pragma once
-#include <algorithm>
-#include <array>
-#include <bit>
-#include <memory>
-#include <mutex>
-#include <optional>
-#include <span>
-#include <type_traits>
-#include <unordered_map>
-#include <unordered_set>
-#include <utility>
-#include <vector>
-
-#include <boost/container/small_vector.hpp>
-
#include "common/alignment.h"
-#include "common/common_types.h"
-#include "common/literals.h"
-#include "common/logging/log.h"
#include "common/settings.h"
-#include "video_core/compatible_formats.h"
-#include "video_core/delayed_destruction_ring.h"
#include "video_core/dirty_flags.h"
-#include "video_core/engines/fermi_2d.h"
-#include "video_core/engines/kepler_compute.h"
-#include "video_core/engines/maxwell_3d.h"
-#include "video_core/memory_manager.h"
-#include "video_core/rasterizer_interface.h"
-#include "video_core/surface.h"
-#include "video_core/texture_cache/descriptor_table.h"
-#include "video_core/texture_cache/format_lookup_table.h"
-#include "video_core/texture_cache/formatter.h"
-#include "video_core/texture_cache/image_base.h"
-#include "video_core/texture_cache/image_info.h"
-#include "video_core/texture_cache/image_view_base.h"
-#include "video_core/texture_cache/image_view_info.h"
-#include "video_core/texture_cache/render_targets.h"
#include "video_core/texture_cache/samples_helper.h"
-#include "video_core/texture_cache/slot_vector.h"
-#include "video_core/texture_cache/types.h"
-#include "video_core/texture_cache/util.h"
-#include "video_core/textures/texture.h"
+#include "video_core/texture_cache/texture_cache_base.h"
namespace VideoCommon {
@@ -62,352 +25,6 @@ using VideoCore::Surface::SurfaceType;
using namespace Common::Literals;
template <class P>
-class TextureCache {
- /// Address shift for caching images into a hash table
- static constexpr u64 PAGE_BITS = 20;
-
- /// Enables debugging features to the texture cache
- static constexpr bool ENABLE_VALIDATION = P::ENABLE_VALIDATION;
- /// Implement blits as copies between framebuffers
- static constexpr bool FRAMEBUFFER_BLITS = P::FRAMEBUFFER_BLITS;
- /// True when some copies have to be emulated
- static constexpr bool HAS_EMULATED_COPIES = P::HAS_EMULATED_COPIES;
- /// True when the API can provide info about the memory of the device.
- static constexpr bool HAS_DEVICE_MEMORY_INFO = P::HAS_DEVICE_MEMORY_INFO;
-
- /// Image view ID for null descriptors
- static constexpr ImageViewId NULL_IMAGE_VIEW_ID{0};
- /// Sampler ID for bugged sampler ids
- static constexpr SamplerId NULL_SAMPLER_ID{0};
-
- static constexpr u64 DEFAULT_EXPECTED_MEMORY = 1_GiB;
- static constexpr u64 DEFAULT_CRITICAL_MEMORY = 2_GiB;
-
- using Runtime = typename P::Runtime;
- using Image = typename P::Image;
- using ImageAlloc = typename P::ImageAlloc;
- using ImageView = typename P::ImageView;
- using Sampler = typename P::Sampler;
- using Framebuffer = typename P::Framebuffer;
-
- struct BlitImages {
- ImageId dst_id;
- ImageId src_id;
- PixelFormat dst_format;
- PixelFormat src_format;
- };
-
- template <typename T>
- struct IdentityHash {
- [[nodiscard]] size_t operator()(T value) const noexcept {
- return static_cast<size_t>(value);
- }
- };
-
-public:
- explicit TextureCache(Runtime&, VideoCore::RasterizerInterface&, Tegra::Engines::Maxwell3D&,
- Tegra::Engines::KeplerCompute&, Tegra::MemoryManager&);
-
- /// Notify the cache that a new frame has been queued
- void TickFrame();
-
- /// Return a constant reference to the given image view id
- [[nodiscard]] const ImageView& GetImageView(ImageViewId id) const noexcept;
-
- /// Return a reference to the given image view id
- [[nodiscard]] ImageView& GetImageView(ImageViewId id) noexcept;
-
- /// Mark an image as modified from the GPU
- void MarkModification(ImageId id) noexcept;
-
- /// Fill image_view_ids with the graphics images in indices
- void FillGraphicsImageViews(std::span<const u32> indices,
- std::span<ImageViewId> image_view_ids);
-
- /// Fill image_view_ids with the compute images in indices
- void FillComputeImageViews(std::span<const u32> indices, std::span<ImageViewId> image_view_ids);
-
- /// Get the sampler from the graphics descriptor table in the specified index
- Sampler* GetGraphicsSampler(u32 index);
-
- /// Get the sampler from the compute descriptor table in the specified index
- Sampler* GetComputeSampler(u32 index);
-
- /// Refresh the state for graphics image view and sampler descriptors
- void SynchronizeGraphicsDescriptors();
-
- /// Refresh the state for compute image view and sampler descriptors
- void SynchronizeComputeDescriptors();
-
- /// Update bound render targets and upload memory if necessary
- /// @param is_clear True when the render targets are being used for clears
- void UpdateRenderTargets(bool is_clear);
-
- /// Find a framebuffer with the currently bound render targets
- /// UpdateRenderTargets should be called before this
- Framebuffer* GetFramebuffer();
-
- /// Mark images in a range as modified from the CPU
- void WriteMemory(VAddr cpu_addr, size_t size);
-
- /// Download contents of host images to guest memory in a region
- void DownloadMemory(VAddr cpu_addr, size_t size);
-
- /// Remove images in a region
- void UnmapMemory(VAddr cpu_addr, size_t size);
-
- /// Remove images in a region
- void UnmapGPUMemory(GPUVAddr gpu_addr, size_t size);
-
- /// Blit an image with the given parameters
- void BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
- const Tegra::Engines::Fermi2D::Surface& src,
- const Tegra::Engines::Fermi2D::Config& copy);
-
- /// Invalidate the contents of the color buffer index
- /// These contents become unspecified, the cache can assume aggressive optimizations.
- void InvalidateColorBuffer(size_t index);
-
- /// Invalidate the contents of the depth buffer
- /// These contents become unspecified, the cache can assume aggressive optimizations.
- void InvalidateDepthBuffer();
-
- /// Try to find a cached image view in the given CPU address
- [[nodiscard]] ImageView* TryFindFramebufferImageView(VAddr cpu_addr);
-
- /// Return true when there are uncommitted images to be downloaded
- [[nodiscard]] bool HasUncommittedFlushes() const noexcept;
-
- /// Return true when the caller should wait for async downloads
- [[nodiscard]] bool ShouldWaitAsyncFlushes() const noexcept;
-
- /// Commit asynchronous downloads
- void CommitAsyncFlushes();
-
- /// Pop asynchronous downloads
- void PopAsyncFlushes();
-
- /// Return true when a CPU region is modified from the GPU
- [[nodiscard]] bool IsRegionGpuModified(VAddr addr, size_t size);
-
- std::mutex mutex;
-
-private:
- /// Iterate over all page indices in a range
- template <typename Func>
- static void ForEachCPUPage(VAddr addr, size_t size, Func&& func) {
- static constexpr bool RETURNS_BOOL = std::is_same_v<std::invoke_result<Func, u64>, bool>;
- const u64 page_end = (addr + size - 1) >> PAGE_BITS;
- for (u64 page = addr >> PAGE_BITS; page <= page_end; ++page) {
- if constexpr (RETURNS_BOOL) {
- if (func(page)) {
- break;
- }
- } else {
- func(page);
- }
- }
- }
-
- template <typename Func>
- static void ForEachGPUPage(GPUVAddr addr, size_t size, Func&& func) {
- static constexpr bool RETURNS_BOOL = std::is_same_v<std::invoke_result<Func, u64>, bool>;
- const u64 page_end = (addr + size - 1) >> PAGE_BITS;
- for (u64 page = addr >> PAGE_BITS; page <= page_end; ++page) {
- if constexpr (RETURNS_BOOL) {
- if (func(page)) {
- break;
- }
- } else {
- func(page);
- }
- }
- }
-
- /// Runs the Garbage Collector.
- void RunGarbageCollector();
-
- /// Fills image_view_ids in the image views in indices
- void FillImageViews(DescriptorTable<TICEntry>& table,
- std::span<ImageViewId> cached_image_view_ids, std::span<const u32> indices,
- std::span<ImageViewId> image_view_ids);
-
- /// Find or create an image view in the guest descriptor table
- ImageViewId VisitImageView(DescriptorTable<TICEntry>& table,
- std::span<ImageViewId> cached_image_view_ids, u32 index);
-
- /// Find or create a framebuffer with the given render target parameters
- FramebufferId GetFramebufferId(const RenderTargets& key);
-
- /// Refresh the contents (pixel data) of an image
- void RefreshContents(Image& image, ImageId image_id);
-
- /// Upload data from guest to an image
- template <typename StagingBuffer>
- void UploadImageContents(Image& image, StagingBuffer& staging_buffer);
-
- /// Find or create an image view from a guest descriptor
- [[nodiscard]] ImageViewId FindImageView(const TICEntry& config);
-
- /// Create a new image view from a guest descriptor
- [[nodiscard]] ImageViewId CreateImageView(const TICEntry& config);
-
- /// Find or create an image from the given parameters
- [[nodiscard]] ImageId FindOrInsertImage(const ImageInfo& info, GPUVAddr gpu_addr,
- RelaxedOptions options = RelaxedOptions{});
-
- /// Find an image from the given parameters
- [[nodiscard]] ImageId FindImage(const ImageInfo& info, GPUVAddr gpu_addr,
- RelaxedOptions options);
-
- /// Create an image from the given parameters
- [[nodiscard]] ImageId InsertImage(const ImageInfo& info, GPUVAddr gpu_addr,
- RelaxedOptions options);
-
- /// Create a new image and join perfectly matching existing images
- /// Remove joined images from the cache
- [[nodiscard]] ImageId JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VAddr cpu_addr);
-
- /// Return a blit image pair from the given guest blit parameters
- [[nodiscard]] BlitImages GetBlitImages(const Tegra::Engines::Fermi2D::Surface& dst,
- const Tegra::Engines::Fermi2D::Surface& src);
-
- /// Find or create a sampler from a guest descriptor sampler
- [[nodiscard]] SamplerId FindSampler(const TSCEntry& config);
-
- /// Find or create an image view for the given color buffer index
- [[nodiscard]] ImageViewId FindColorBuffer(size_t index, bool is_clear);
-
- /// Find or create an image view for the depth buffer
- [[nodiscard]] ImageViewId FindDepthBuffer(bool is_clear);
-
- /// Find or create a view for a render target with the given image parameters
- [[nodiscard]] ImageViewId FindRenderTargetView(const ImageInfo& info, GPUVAddr gpu_addr,
- bool is_clear);
-
- /// Iterates over all the images in a region calling func
- template <typename Func>
- void ForEachImageInRegion(VAddr cpu_addr, size_t size, Func&& func);
-
- template <typename Func>
- void ForEachImageInRegionGPU(GPUVAddr gpu_addr, size_t size, Func&& func);
-
- template <typename Func>
- void ForEachSparseImageInRegion(GPUVAddr gpu_addr, size_t size, Func&& func);
-
- /// Iterates over all the images in a region calling func
- template <typename Func>
- void ForEachSparseSegment(ImageBase& image, Func&& func);
-
- /// Find or create an image view in the given image with the passed parameters
- [[nodiscard]] ImageViewId FindOrEmplaceImageView(ImageId image_id, const ImageViewInfo& info);
-
- /// Register image in the page table
- void RegisterImage(ImageId image);
-
- /// Unregister image from the page table
- void UnregisterImage(ImageId image);
-
- /// Track CPU reads and writes for image
- void TrackImage(ImageBase& image, ImageId image_id);
-
- /// Stop tracking CPU reads and writes for image
- void UntrackImage(ImageBase& image, ImageId image_id);
-
- /// Delete image from the cache
- void DeleteImage(ImageId image);
-
- /// Remove image views references from the cache
- void RemoveImageViewReferences(std::span<const ImageViewId> removed_views);
-
- /// Remove framebuffers using the given image views from the cache
- void RemoveFramebuffers(std::span<const ImageViewId> removed_views);
-
- /// Mark an image as modified from the GPU
- void MarkModification(ImageBase& image) noexcept;
-
- /// Synchronize image aliases, copying data if needed
- void SynchronizeAliases(ImageId image_id);
-
- /// Prepare an image to be used
- void PrepareImage(ImageId image_id, bool is_modification, bool invalidate);
-
- /// Prepare an image view to be used
- void PrepareImageView(ImageViewId image_view_id, bool is_modification, bool invalidate);
-
- /// Execute copies from one image to the other, even if they are incompatible
- void CopyImage(ImageId dst_id, ImageId src_id, std::span<const ImageCopy> copies);
-
- /// Bind an image view as render target, downloading resources preemtively if needed
- void BindRenderTarget(ImageViewId* old_id, ImageViewId new_id);
-
- /// Create a render target from a given image and image view parameters
- [[nodiscard]] std::pair<FramebufferId, ImageViewId> RenderTargetFromImage(
- ImageId, const ImageViewInfo& view_info);
-
- /// Returns true if the current clear parameters clear the whole image of a given image view
- [[nodiscard]] bool IsFullClear(ImageViewId id);
-
- Runtime& runtime;
- VideoCore::RasterizerInterface& rasterizer;
- Tegra::Engines::Maxwell3D& maxwell3d;
- Tegra::Engines::KeplerCompute& kepler_compute;
- Tegra::MemoryManager& gpu_memory;
-
- DescriptorTable<TICEntry> graphics_image_table{gpu_memory};
- DescriptorTable<TSCEntry> graphics_sampler_table{gpu_memory};
- std::vector<SamplerId> graphics_sampler_ids;
- std::vector<ImageViewId> graphics_image_view_ids;
-
- DescriptorTable<TICEntry> compute_image_table{gpu_memory};
- DescriptorTable<TSCEntry> compute_sampler_table{gpu_memory};
- std::vector<SamplerId> compute_sampler_ids;
- std::vector<ImageViewId> compute_image_view_ids;
-
- RenderTargets render_targets;
-
- std::unordered_map<TICEntry, ImageViewId> image_views;
- std::unordered_map<TSCEntry, SamplerId> samplers;
- std::unordered_map<RenderTargets, FramebufferId> framebuffers;
-
- std::unordered_map<u64, std::vector<ImageMapId>, IdentityHash<u64>> page_table;
- std::unordered_map<u64, std::vector<ImageId>, IdentityHash<u64>> gpu_page_table;
- std::unordered_map<u64, std::vector<ImageId>, IdentityHash<u64>> sparse_page_table;
-
- std::unordered_map<ImageId, std::vector<ImageViewId>> sparse_views;
-
- VAddr virtual_invalid_space{};
-
- bool has_deleted_images = false;
- u64 total_used_memory = 0;
- u64 minimum_memory;
- u64 expected_memory;
- u64 critical_memory;
-
- SlotVector<Image> slot_images;
- SlotVector<ImageMapView> slot_map_views;
- SlotVector<ImageView> slot_image_views;
- SlotVector<ImageAlloc> slot_image_allocs;
- SlotVector<Sampler> slot_samplers;
- SlotVector<Framebuffer> slot_framebuffers;
-
- // TODO: This data structure is not optimal and it should be reworked
- std::vector<ImageId> uncommitted_downloads;
- std::queue<std::vector<ImageId>> committed_downloads;
-
- static constexpr size_t TICKS_TO_DESTROY = 6;
- DelayedDestructionRing<Image, TICKS_TO_DESTROY> sentenced_images;
- DelayedDestructionRing<ImageView, TICKS_TO_DESTROY> sentenced_image_view;
- DelayedDestructionRing<Framebuffer, TICKS_TO_DESTROY> sentenced_framebuffers;
-
- std::unordered_map<GPUVAddr, ImageAllocId> image_allocs_table;
-
- u64 modification_tick = 0;
- u64 frame_tick = 0;
- typename SlotVector<Image>::Iterator deletion_iterator;
-};
-
-template <class P>
TextureCache<P>::TextureCache(Runtime& runtime_, VideoCore::RasterizerInterface& rasterizer_,
Tegra::Engines::Maxwell3D& maxwell3d_,
Tegra::Engines::KeplerCompute& kepler_compute_,
@@ -821,40 +438,6 @@ void TextureCache<P>::BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
}
template <class P>
-void TextureCache<P>::InvalidateColorBuffer(size_t index) {
- ImageViewId& color_buffer_id = render_targets.color_buffer_ids[index];
- color_buffer_id = FindColorBuffer(index, false);
- if (!color_buffer_id) {
- LOG_ERROR(HW_GPU, "Invalidating invalid color buffer in index={}", index);
- return;
- }
- // When invalidating a color buffer, the old contents are no longer relevant
- ImageView& color_buffer = slot_image_views[color_buffer_id];
- Image& image = slot_images[color_buffer.image_id];
- image.flags &= ~ImageFlagBits::CpuModified;
- image.flags &= ~ImageFlagBits::GpuModified;
-
- runtime.InvalidateColorBuffer(color_buffer, index);
-}
-
-template <class P>
-void TextureCache<P>::InvalidateDepthBuffer() {
- ImageViewId& depth_buffer_id = render_targets.depth_buffer_id;
- depth_buffer_id = FindDepthBuffer(false);
- if (!depth_buffer_id) {
- LOG_ERROR(HW_GPU, "Invalidating invalid depth buffer");
- return;
- }
- // When invalidating the depth buffer, the old contents are no longer relevant
- ImageBase& image = slot_images[slot_image_views[depth_buffer_id].image_id];
- image.flags &= ~ImageFlagBits::CpuModified;
- image.flags &= ~ImageFlagBits::GpuModified;
-
- ImageView& depth_buffer = slot_image_views[depth_buffer_id];
- runtime.InvalidateDepthBuffer(depth_buffer);
-}
-
-template <class P>
typename P::ImageView* TextureCache<P>::TryFindFramebufferImageView(VAddr cpu_addr) {
// TODO: Properly implement this
const auto it = page_table.find(cpu_addr >> PAGE_BITS);
diff --git a/src/video_core/texture_cache/texture_cache_base.h b/src/video_core/texture_cache/texture_cache_base.h
new file mode 100644
index 000000000..e4ae351cb
--- /dev/null
+++ b/src/video_core/texture_cache/texture_cache_base.h
@@ -0,0 +1,385 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <mutex>
+#include <span>
+#include <type_traits>
+#include <unordered_map>
+#include <unordered_set>
+#include <vector>
+
+#include "common/common_types.h"
+#include "common/literals.h"
+#include "video_core/compatible_formats.h"
+#include "video_core/delayed_destruction_ring.h"
+#include "video_core/engines/fermi_2d.h"
+#include "video_core/engines/kepler_compute.h"
+#include "video_core/engines/maxwell_3d.h"
+#include "video_core/memory_manager.h"
+#include "video_core/rasterizer_interface.h"
+#include "video_core/surface.h"
+#include "video_core/texture_cache/descriptor_table.h"
+#include "video_core/texture_cache/image_base.h"
+#include "video_core/texture_cache/image_info.h"
+#include "video_core/texture_cache/image_view_info.h"
+#include "video_core/texture_cache/render_targets.h"
+#include "video_core/texture_cache/slot_vector.h"
+#include "video_core/texture_cache/types.h"
+#include "video_core/texture_cache/util.h"
+#include "video_core/textures/texture.h"
+
+namespace VideoCommon {
+
+using Tegra::Texture::SwizzleSource;
+using Tegra::Texture::TICEntry;
+using Tegra::Texture::TSCEntry;
+using VideoCore::Surface::GetFormatType;
+using VideoCore::Surface::IsCopyCompatible;
+using VideoCore::Surface::PixelFormat;
+using VideoCore::Surface::PixelFormatFromDepthFormat;
+using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
+using namespace Common::Literals;
+
+template <class P>
+class TextureCache {
+ /// Address shift for caching images into a hash table
+ static constexpr u64 PAGE_BITS = 20;
+
+ /// Enables debugging features to the texture cache
+ static constexpr bool ENABLE_VALIDATION = P::ENABLE_VALIDATION;
+ /// Implement blits as copies between framebuffers
+ static constexpr bool FRAMEBUFFER_BLITS = P::FRAMEBUFFER_BLITS;
+ /// True when some copies have to be emulated
+ static constexpr bool HAS_EMULATED_COPIES = P::HAS_EMULATED_COPIES;
+ /// True when the API can provide info about the memory of the device.
+ static constexpr bool HAS_DEVICE_MEMORY_INFO = P::HAS_DEVICE_MEMORY_INFO;
+
+ /// Image view ID for null descriptors
+ static constexpr ImageViewId NULL_IMAGE_VIEW_ID{0};
+ /// Sampler ID for bugged sampler ids
+ static constexpr SamplerId NULL_SAMPLER_ID{0};
+
+ static constexpr u64 DEFAULT_EXPECTED_MEMORY = 1_GiB;
+ static constexpr u64 DEFAULT_CRITICAL_MEMORY = 2_GiB;
+
+ using Runtime = typename P::Runtime;
+ using Image = typename P::Image;
+ using ImageAlloc = typename P::ImageAlloc;
+ using ImageView = typename P::ImageView;
+ using Sampler = typename P::Sampler;
+ using Framebuffer = typename P::Framebuffer;
+
+ struct BlitImages {
+ ImageId dst_id;
+ ImageId src_id;
+ PixelFormat dst_format;
+ PixelFormat src_format;
+ };
+
+ template <typename T>
+ struct IdentityHash {
+ [[nodiscard]] size_t operator()(T value) const noexcept {
+ return static_cast<size_t>(value);
+ }
+ };
+
+public:
+ explicit TextureCache(Runtime&, VideoCore::RasterizerInterface&, Tegra::Engines::Maxwell3D&,
+ Tegra::Engines::KeplerCompute&, Tegra::MemoryManager&);
+
+ /// Notify the cache that a new frame has been queued
+ void TickFrame();
+
+ /// Return a constant reference to the given image view id
+ [[nodiscard]] const ImageView& GetImageView(ImageViewId id) const noexcept;
+
+ /// Return a reference to the given image view id
+ [[nodiscard]] ImageView& GetImageView(ImageViewId id) noexcept;
+
+ /// Mark an image as modified from the GPU
+ void MarkModification(ImageId id) noexcept;
+
+ /// Fill image_view_ids with the graphics images in indices
+ void FillGraphicsImageViews(std::span<const u32> indices,
+ std::span<ImageViewId> image_view_ids);
+
+ /// Fill image_view_ids with the compute images in indices
+ void FillComputeImageViews(std::span<const u32> indices, std::span<ImageViewId> image_view_ids);
+
+ /// Get the sampler from the graphics descriptor table in the specified index
+ Sampler* GetGraphicsSampler(u32 index);
+
+ /// Get the sampler from the compute descriptor table in the specified index
+ Sampler* GetComputeSampler(u32 index);
+
+ /// Refresh the state for graphics image view and sampler descriptors
+ void SynchronizeGraphicsDescriptors();
+
+ /// Refresh the state for compute image view and sampler descriptors
+ void SynchronizeComputeDescriptors();
+
+ /// Update bound render targets and upload memory if necessary
+ /// @param is_clear True when the render targets are being used for clears
+ void UpdateRenderTargets(bool is_clear);
+
+ /// Find a framebuffer with the currently bound render targets
+ /// UpdateRenderTargets should be called before this
+ Framebuffer* GetFramebuffer();
+
+ /// Mark images in a range as modified from the CPU
+ void WriteMemory(VAddr cpu_addr, size_t size);
+
+ /// Download contents of host images to guest memory in a region
+ void DownloadMemory(VAddr cpu_addr, size_t size);
+
+ /// Remove images in a region
+ void UnmapMemory(VAddr cpu_addr, size_t size);
+
+ /// Remove images in a region
+ void UnmapGPUMemory(GPUVAddr gpu_addr, size_t size);
+
+ /// Blit an image with the given parameters
+ void BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
+ const Tegra::Engines::Fermi2D::Surface& src,
+ const Tegra::Engines::Fermi2D::Config& copy);
+
+ /// Try to find a cached image view in the given CPU address
+ [[nodiscard]] ImageView* TryFindFramebufferImageView(VAddr cpu_addr);
+
+ /// Return true when there are uncommitted images to be downloaded
+ [[nodiscard]] bool HasUncommittedFlushes() const noexcept;
+
+ /// Return true when the caller should wait for async downloads
+ [[nodiscard]] bool ShouldWaitAsyncFlushes() const noexcept;
+
+ /// Commit asynchronous downloads
+ void CommitAsyncFlushes();
+
+ /// Pop asynchronous downloads
+ void PopAsyncFlushes();
+
+ /// Return true when a CPU region is modified from the GPU
+ [[nodiscard]] bool IsRegionGpuModified(VAddr addr, size_t size);
+
+ std::mutex mutex;
+
+private:
+ /// Iterate over all page indices in a range
+ template <typename Func>
+ static void ForEachCPUPage(VAddr addr, size_t size, Func&& func) {
+ static constexpr bool RETURNS_BOOL = std::is_same_v<std::invoke_result<Func, u64>, bool>;
+ const u64 page_end = (addr + size - 1) >> PAGE_BITS;
+ for (u64 page = addr >> PAGE_BITS; page <= page_end; ++page) {
+ if constexpr (RETURNS_BOOL) {
+ if (func(page)) {
+ break;
+ }
+ } else {
+ func(page);
+ }
+ }
+ }
+
+ template <typename Func>
+ static void ForEachGPUPage(GPUVAddr addr, size_t size, Func&& func) {
+ static constexpr bool RETURNS_BOOL = std::is_same_v<std::invoke_result<Func, u64>, bool>;
+ const u64 page_end = (addr + size - 1) >> PAGE_BITS;
+ for (u64 page = addr >> PAGE_BITS; page <= page_end; ++page) {
+ if constexpr (RETURNS_BOOL) {
+ if (func(page)) {
+ break;
+ }
+ } else {
+ func(page);
+ }
+ }
+ }
+
+ /// Runs the Garbage Collector.
+ void RunGarbageCollector();
+
+ /// Fills image_view_ids in the image views in indices
+ void FillImageViews(DescriptorTable<TICEntry>& table,
+ std::span<ImageViewId> cached_image_view_ids, std::span<const u32> indices,
+ std::span<ImageViewId> image_view_ids);
+
+ /// Find or create an image view in the guest descriptor table
+ ImageViewId VisitImageView(DescriptorTable<TICEntry>& table,
+ std::span<ImageViewId> cached_image_view_ids, u32 index);
+
+ /// Find or create a framebuffer with the given render target parameters
+ FramebufferId GetFramebufferId(const RenderTargets& key);
+
+ /// Refresh the contents (pixel data) of an image
+ void RefreshContents(Image& image, ImageId image_id);
+
+ /// Upload data from guest to an image
+ template <typename StagingBuffer>
+ void UploadImageContents(Image& image, StagingBuffer& staging_buffer);
+
+ /// Find or create an image view from a guest descriptor
+ [[nodiscard]] ImageViewId FindImageView(const TICEntry& config);
+
+ /// Create a new image view from a guest descriptor
+ [[nodiscard]] ImageViewId CreateImageView(const TICEntry& config);
+
+ /// Find or create an image from the given parameters
+ [[nodiscard]] ImageId FindOrInsertImage(const ImageInfo& info, GPUVAddr gpu_addr,
+ RelaxedOptions options = RelaxedOptions{});
+
+ /// Find an image from the given parameters
+ [[nodiscard]] ImageId FindImage(const ImageInfo& info, GPUVAddr gpu_addr,
+ RelaxedOptions options);
+
+ /// Create an image from the given parameters
+ [[nodiscard]] ImageId InsertImage(const ImageInfo& info, GPUVAddr gpu_addr,
+ RelaxedOptions options);
+
+ /// Create a new image and join perfectly matching existing images
+ /// Remove joined images from the cache
+ [[nodiscard]] ImageId JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VAddr cpu_addr);
+
+ /// Return a blit image pair from the given guest blit parameters
+ [[nodiscard]] BlitImages GetBlitImages(const Tegra::Engines::Fermi2D::Surface& dst,
+ const Tegra::Engines::Fermi2D::Surface& src);
+
+ /// Find or create a sampler from a guest descriptor sampler
+ [[nodiscard]] SamplerId FindSampler(const TSCEntry& config);
+
+ /// Find or create an image view for the given color buffer index
+ [[nodiscard]] ImageViewId FindColorBuffer(size_t index, bool is_clear);
+
+ /// Find or create an image view for the depth buffer
+ [[nodiscard]] ImageViewId FindDepthBuffer(bool is_clear);
+
+ /// Find or create a view for a render target with the given image parameters
+ [[nodiscard]] ImageViewId FindRenderTargetView(const ImageInfo& info, GPUVAddr gpu_addr,
+ bool is_clear);
+
+ /// Iterates over all the images in a region calling func
+ template <typename Func>
+ void ForEachImageInRegion(VAddr cpu_addr, size_t size, Func&& func);
+
+ template <typename Func>
+ void ForEachImageInRegionGPU(GPUVAddr gpu_addr, size_t size, Func&& func);
+
+ template <typename Func>
+ void ForEachSparseImageInRegion(GPUVAddr gpu_addr, size_t size, Func&& func);
+
+ /// Iterates over all the images in a region calling func
+ template <typename Func>
+ void ForEachSparseSegment(ImageBase& image, Func&& func);
+
+ /// Find or create an image view in the given image with the passed parameters
+ [[nodiscard]] ImageViewId FindOrEmplaceImageView(ImageId image_id, const ImageViewInfo& info);
+
+ /// Register image in the page table
+ void RegisterImage(ImageId image);
+
+ /// Unregister image from the page table
+ void UnregisterImage(ImageId image);
+
+ /// Track CPU reads and writes for image
+ void TrackImage(ImageBase& image, ImageId image_id);
+
+ /// Stop tracking CPU reads and writes for image
+ void UntrackImage(ImageBase& image, ImageId image_id);
+
+ /// Delete image from the cache
+ void DeleteImage(ImageId image);
+
+ /// Remove image views references from the cache
+ void RemoveImageViewReferences(std::span<const ImageViewId> removed_views);
+
+ /// Remove framebuffers using the given image views from the cache
+ void RemoveFramebuffers(std::span<const ImageViewId> removed_views);
+
+ /// Mark an image as modified from the GPU
+ void MarkModification(ImageBase& image) noexcept;
+
+ /// Synchronize image aliases, copying data if needed
+ void SynchronizeAliases(ImageId image_id);
+
+ /// Prepare an image to be used
+ void PrepareImage(ImageId image_id, bool is_modification, bool invalidate);
+
+ /// Prepare an image view to be used
+ void PrepareImageView(ImageViewId image_view_id, bool is_modification, bool invalidate);
+
+ /// Execute copies from one image to the other, even if they are incompatible
+ void CopyImage(ImageId dst_id, ImageId src_id, std::span<const ImageCopy> copies);
+
+ /// Bind an image view as render target, downloading resources preemtively if needed
+ void BindRenderTarget(ImageViewId* old_id, ImageViewId new_id);
+
+ /// Create a render target from a given image and image view parameters
+ [[nodiscard]] std::pair<FramebufferId, ImageViewId> RenderTargetFromImage(
+ ImageId, const ImageViewInfo& view_info);
+
+ /// Returns true if the current clear parameters clear the whole image of a given image view
+ [[nodiscard]] bool IsFullClear(ImageViewId id);
+
+ Runtime& runtime;
+ VideoCore::RasterizerInterface& rasterizer;
+ Tegra::Engines::Maxwell3D& maxwell3d;
+ Tegra::Engines::KeplerCompute& kepler_compute;
+ Tegra::MemoryManager& gpu_memory;
+
+ DescriptorTable<TICEntry> graphics_image_table{gpu_memory};
+ DescriptorTable<TSCEntry> graphics_sampler_table{gpu_memory};
+ std::vector<SamplerId> graphics_sampler_ids;
+ std::vector<ImageViewId> graphics_image_view_ids;
+
+ DescriptorTable<TICEntry> compute_image_table{gpu_memory};
+ DescriptorTable<TSCEntry> compute_sampler_table{gpu_memory};
+ std::vector<SamplerId> compute_sampler_ids;
+ std::vector<ImageViewId> compute_image_view_ids;
+
+ RenderTargets render_targets;
+
+ std::unordered_map<TICEntry, ImageViewId> image_views;
+ std::unordered_map<TSCEntry, SamplerId> samplers;
+ std::unordered_map<RenderTargets, FramebufferId> framebuffers;
+
+ std::unordered_map<u64, std::vector<ImageMapId>, IdentityHash<u64>> page_table;
+ std::unordered_map<u64, std::vector<ImageId>, IdentityHash<u64>> gpu_page_table;
+ std::unordered_map<u64, std::vector<ImageId>, IdentityHash<u64>> sparse_page_table;
+
+ std::unordered_map<ImageId, std::vector<ImageViewId>> sparse_views;
+
+ VAddr virtual_invalid_space{};
+
+ bool has_deleted_images = false;
+ u64 total_used_memory = 0;
+ u64 minimum_memory;
+ u64 expected_memory;
+ u64 critical_memory;
+
+ SlotVector<Image> slot_images;
+ SlotVector<ImageMapView> slot_map_views;
+ SlotVector<ImageView> slot_image_views;
+ SlotVector<ImageAlloc> slot_image_allocs;
+ SlotVector<Sampler> slot_samplers;
+ SlotVector<Framebuffer> slot_framebuffers;
+
+ // TODO: This data structure is not optimal and it should be reworked
+ std::vector<ImageId> uncommitted_downloads;
+ std::queue<std::vector<ImageId>> committed_downloads;
+
+ static constexpr size_t TICKS_TO_DESTROY = 6;
+ DelayedDestructionRing<Image, TICKS_TO_DESTROY> sentenced_images;
+ DelayedDestructionRing<ImageView, TICKS_TO_DESTROY> sentenced_image_view;
+ DelayedDestructionRing<Framebuffer, TICKS_TO_DESTROY> sentenced_framebuffers;
+
+ std::unordered_map<GPUVAddr, ImageAllocId> image_allocs_table;
+
+ u64 modification_tick = 0;
+ u64 frame_tick = 0;
+ typename SlotVector<Image>::Iterator deletion_iterator;
+};
+
+} // namespace VideoCommon