diff options
Diffstat (limited to 'src/video_core/texture_cache')
| -rw-r--r-- | src/video_core/texture_cache/accelerated_swizzle.cpp | 4 | ||||
| -rw-r--r-- | src/video_core/texture_cache/image_base.cpp | 5 | ||||
| -rw-r--r-- | src/video_core/texture_cache/image_view_base.cpp | 2 | ||||
| -rw-r--r-- | src/video_core/texture_cache/texture_cache.h | 73 | ||||
| -rw-r--r-- | src/video_core/texture_cache/util.cpp | 118 | ||||
| -rw-r--r-- | src/video_core/texture_cache/util.h | 9 |
6 files changed, 93 insertions, 118 deletions
diff --git a/src/video_core/texture_cache/accelerated_swizzle.cpp b/src/video_core/texture_cache/accelerated_swizzle.cpp index a4fc1184b..15585caeb 100644 --- a/src/video_core/texture_cache/accelerated_swizzle.cpp +++ b/src/video_core/texture_cache/accelerated_swizzle.cpp @@ -27,7 +27,7 @@ BlockLinearSwizzle2DParams MakeBlockLinearSwizzle2DParams(const SwizzleParameter const Extent3D num_tiles = swizzle.num_tiles; const u32 bytes_per_block = BytesPerBlock(info.format); const u32 stride_alignment = CalculateLevelStrideAlignment(info, swizzle.level); - const u32 stride = Common::AlignBits(num_tiles.width, stride_alignment) * bytes_per_block; + const u32 stride = Common::AlignUpLog2(num_tiles.width, stride_alignment) * bytes_per_block; const u32 gobs_in_x = Common::DivCeilLog2(stride, GOB_SIZE_X_SHIFT); return BlockLinearSwizzle2DParams{ .origin{0, 0, 0}, @@ -47,7 +47,7 @@ BlockLinearSwizzle3DParams MakeBlockLinearSwizzle3DParams(const SwizzleParameter const Extent3D num_tiles = swizzle.num_tiles; const u32 bytes_per_block = BytesPerBlock(info.format); const u32 stride_alignment = CalculateLevelStrideAlignment(info, swizzle.level); - const u32 stride = Common::AlignBits(num_tiles.width, stride_alignment) * bytes_per_block; + const u32 stride = Common::AlignUpLog2(num_tiles.width, stride_alignment) * bytes_per_block; const u32 gobs_in_x = (stride + GOB_SIZE_X - 1) >> GOB_SIZE_X_SHIFT; const u32 block_size = gobs_in_x << (GOB_SIZE_SHIFT + block.height + block.depth); diff --git a/src/video_core/texture_cache/image_base.cpp b/src/video_core/texture_cache/image_base.cpp index 959b3f115..9914926b3 100644 --- a/src/video_core/texture_cache/image_base.cpp +++ b/src/video_core/texture_cache/image_base.cpp @@ -120,9 +120,10 @@ void AddImageAlias(ImageBase& lhs, ImageBase& rhs, ImageId lhs_id, ImageId rhs_i if (lhs.info.type == ImageType::Linear) { base = SubresourceBase{.level = 0, .layer = 0}; } else { - // We are passing relaxed formats as an option, having broken views or not won't matter + // We are passing relaxed formats as an option, having broken views/bgr or not won't matter static constexpr bool broken_views = false; - base = FindSubresource(rhs.info, lhs, rhs.gpu_addr, OPTIONS, broken_views); + static constexpr bool native_bgr = true; + base = FindSubresource(rhs.info, lhs, rhs.gpu_addr, OPTIONS, broken_views, native_bgr); } if (!base) { LOG_ERROR(HW_GPU, "Image alias should have been flipped"); diff --git a/src/video_core/texture_cache/image_view_base.cpp b/src/video_core/texture_cache/image_view_base.cpp index 18f72e508..f89a40b4c 100644 --- a/src/video_core/texture_cache/image_view_base.cpp +++ b/src/video_core/texture_cache/image_view_base.cpp @@ -24,7 +24,7 @@ ImageViewBase::ImageViewBase(const ImageViewInfo& info, const ImageInfo& image_i .height = std::max(image_info.size.height >> range.base.level, 1u), .depth = std::max(image_info.size.depth >> range.base.level, 1u), } { - ASSERT_MSG(VideoCore::Surface::IsViewCompatible(image_info.format, info.format, false), + ASSERT_MSG(VideoCore::Surface::IsViewCompatible(image_info.format, info.format, false, true), "Image view format {} is incompatible with image format {}", info.format, image_info.format); const bool is_async = Settings::values.use_asynchronous_gpu_emulation.GetValue(); diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h index d1080300f..98e33c3a0 100644 --- a/src/video_core/texture_cache/texture_cache.h +++ b/src/video_core/texture_cache/texture_cache.h @@ -103,9 +103,6 @@ public: /// Notify the cache that a new frame has been queued void TickFrame(); - /// Return an unique mutually exclusive lock for the cache - [[nodiscard]] std::unique_lock<std::mutex> AcquireLock(); - /// Return a constant reference to the given image view id [[nodiscard]] const ImageView& GetImageView(ImageViewId id) const noexcept; @@ -179,6 +176,8 @@ public: /// 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> @@ -212,8 +211,8 @@ private: void RefreshContents(Image& image); /// Upload data from guest to an image - template <typename MapBuffer> - void UploadImageContents(Image& image, MapBuffer& map, size_t buffer_offset); + 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); @@ -325,8 +324,6 @@ private: RenderTargets render_targets; - std::mutex mutex; - std::unordered_map<TICEntry, ImageViewId> image_views; std::unordered_map<TSCEntry, SamplerId> samplers; std::unordered_map<RenderTargets, FramebufferId> framebuffers; @@ -386,11 +383,6 @@ void TextureCache<P>::TickFrame() { } template <class P> -std::unique_lock<std::mutex> TextureCache<P>::AcquireLock() { - return std::unique_lock{mutex}; -} - -template <class P> const typename P::ImageView& TextureCache<P>::GetImageView(ImageViewId id) const noexcept { return slot_image_views[id]; } @@ -598,11 +590,11 @@ void TextureCache<P>::DownloadMemory(VAddr cpu_addr, size_t size) { }); for (const ImageId image_id : images) { Image& image = slot_images[image_id]; - auto map = runtime.MapDownloadBuffer(image.unswizzled_size_bytes); + auto map = runtime.DownloadStagingBuffer(image.unswizzled_size_bytes); const auto copies = FullDownloadCopies(image.info); - image.DownloadMemory(map, 0, copies); + image.DownloadMemory(map, copies); runtime.Finish(); - SwizzleImage(gpu_memory, image.gpu_addr, image.info, copies, map.Span()); + SwizzleImage(gpu_memory, image.gpu_addr, image.info, copies, map.mapped_span); } } @@ -757,25 +749,25 @@ void TextureCache<P>::PopAsyncFlushes() { for (const ImageId image_id : download_ids) { total_size_bytes += slot_images[image_id].unswizzled_size_bytes; } - auto download_map = runtime.MapDownloadBuffer(total_size_bytes); - size_t buffer_offset = 0; + auto download_map = runtime.DownloadStagingBuffer(total_size_bytes); + const size_t original_offset = download_map.offset; for (const ImageId image_id : download_ids) { Image& image = slot_images[image_id]; const auto copies = FullDownloadCopies(image.info); - image.DownloadMemory(download_map, buffer_offset, copies); - buffer_offset += image.unswizzled_size_bytes; + image.DownloadMemory(download_map, copies); + download_map.offset += image.unswizzled_size_bytes; } // Wait for downloads to finish runtime.Finish(); - buffer_offset = 0; - const std::span<u8> download_span = download_map.Span(); + download_map.offset = original_offset; + std::span<u8> download_span = download_map.mapped_span; for (const ImageId image_id : download_ids) { const ImageBase& image = slot_images[image_id]; const auto copies = FullDownloadCopies(image.info); - const std::span<u8> image_download_span = download_span.subspan(buffer_offset); - SwizzleImage(gpu_memory, image.gpu_addr, image.info, copies, image_download_span); - buffer_offset += image.unswizzled_size_bytes; + SwizzleImage(gpu_memory, image.gpu_addr, image.info, copies, download_span); + download_map.offset += image.unswizzled_size_bytes; + download_span = download_span.subspan(image.unswizzled_size_bytes); } committed_downloads.pop(); } @@ -806,32 +798,32 @@ void TextureCache<P>::RefreshContents(Image& image) { LOG_WARNING(HW_GPU, "MSAA image uploads are not implemented"); return; } - auto map = runtime.MapUploadBuffer(MapSizeBytes(image)); - UploadImageContents(image, map, 0); + auto staging = runtime.UploadStagingBuffer(MapSizeBytes(image)); + UploadImageContents(image, staging); runtime.InsertUploadMemoryBarrier(); } template <class P> -template <typename MapBuffer> -void TextureCache<P>::UploadImageContents(Image& image, MapBuffer& map, size_t buffer_offset) { - const std::span<u8> mapped_span = map.Span().subspan(buffer_offset); +template <typename StagingBuffer> +void TextureCache<P>::UploadImageContents(Image& image, StagingBuffer& staging) { + const std::span<u8> mapped_span = staging.mapped_span; const GPUVAddr gpu_addr = image.gpu_addr; if (True(image.flags & ImageFlagBits::AcceleratedUpload)) { gpu_memory.ReadBlockUnsafe(gpu_addr, mapped_span.data(), mapped_span.size_bytes()); const auto uploads = FullUploadSwizzles(image.info); - runtime.AccelerateImageUpload(image, map, buffer_offset, uploads); + runtime.AccelerateImageUpload(image, staging, uploads); } else if (True(image.flags & ImageFlagBits::Converted)) { std::vector<u8> unswizzled_data(image.unswizzled_size_bytes); auto copies = UnswizzleImage(gpu_memory, gpu_addr, image.info, unswizzled_data); ConvertImage(unswizzled_data, image.info, mapped_span, copies); - image.UploadMemory(map, buffer_offset, copies); + image.UploadMemory(staging, copies); } else if (image.info.type == ImageType::Buffer) { const std::array copies{UploadBufferCopy(gpu_memory, gpu_addr, image, mapped_span)}; - image.UploadMemory(map, buffer_offset, copies); + image.UploadMemory(staging, copies); } else { const auto copies = UnswizzleImage(gpu_memory, gpu_addr, image.info, mapped_span); - image.UploadMemory(map, buffer_offset, copies); + image.UploadMemory(staging, copies); } } @@ -884,6 +876,7 @@ ImageId TextureCache<P>::FindImage(const ImageInfo& info, GPUVAddr gpu_addr, return ImageId{}; } const bool broken_views = runtime.HasBrokenTextureViewFormats(); + const bool native_bgr = runtime.HasNativeBgr(); ImageId image_id; const auto lambda = [&](ImageId existing_image_id, ImageBase& existing_image) { if (info.type == ImageType::Linear || existing_image.info.type == ImageType::Linear) { @@ -893,11 +886,12 @@ ImageId TextureCache<P>::FindImage(const ImageInfo& info, GPUVAddr gpu_addr, if (existing_image.gpu_addr == gpu_addr && existing.type == info.type && existing.pitch == info.pitch && IsPitchLinearSameSize(existing, info, strict_size) && - IsViewCompatible(existing.format, info.format, broken_views)) { + IsViewCompatible(existing.format, info.format, broken_views, native_bgr)) { image_id = existing_image_id; return true; } - } else if (IsSubresource(info, existing_image, gpu_addr, options, broken_views)) { + } else if (IsSubresource(info, existing_image, gpu_addr, options, broken_views, + native_bgr)) { image_id = existing_image_id; return true; } @@ -928,6 +922,7 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA ImageInfo new_info = info; const size_t size_bytes = CalculateGuestSizeInBytes(new_info); const bool broken_views = runtime.HasBrokenTextureViewFormats(); + const bool native_bgr = runtime.HasNativeBgr(); std::vector<ImageId> overlap_ids; std::vector<ImageId> left_aliased_ids; std::vector<ImageId> right_aliased_ids; @@ -943,8 +938,8 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA return; } static constexpr bool strict_size = true; - const std::optional<OverlapResult> solution = - ResolveOverlap(new_info, gpu_addr, cpu_addr, overlap, strict_size, broken_views); + const std::optional<OverlapResult> solution = ResolveOverlap( + new_info, gpu_addr, cpu_addr, overlap, strict_size, broken_views, native_bgr); if (solution) { gpu_addr = solution->gpu_addr; cpu_addr = solution->cpu_addr; @@ -954,10 +949,10 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA } static constexpr auto options = RelaxedOptions::Size | RelaxedOptions::Format; const ImageBase new_image_base(new_info, gpu_addr, cpu_addr); - if (IsSubresource(new_info, overlap, gpu_addr, options, broken_views)) { + if (IsSubresource(new_info, overlap, gpu_addr, options, broken_views, native_bgr)) { left_aliased_ids.push_back(overlap_id); } else if (IsSubresource(overlap.info, new_image_base, overlap.gpu_addr, options, - broken_views)) { + broken_views, native_bgr)) { right_aliased_ids.push_back(overlap_id); } }); diff --git a/src/video_core/texture_cache/util.cpp b/src/video_core/texture_cache/util.cpp index 279932778..2c42d1449 100644 --- a/src/video_core/texture_cache/util.cpp +++ b/src/video_core/texture_cache/util.cpp @@ -169,40 +169,6 @@ template <u32 GOB_EXTENT> return Common::DivCeil(AdjustMipSize(size, level), block_size); } -[[nodiscard]] constexpr u32 LayerSize(const TICEntry& config, PixelFormat format) { - return config.Width() * config.Height() * BytesPerBlock(format); -} - -[[nodiscard]] constexpr bool HasTwoDimsPerLayer(TextureType type) { - switch (type) { - case TextureType::Texture2D: - case TextureType::Texture2DArray: - case TextureType::Texture2DNoMipmap: - case TextureType::Texture3D: - case TextureType::TextureCubeArray: - case TextureType::TextureCubemap: - return true; - case TextureType::Texture1D: - case TextureType::Texture1DArray: - case TextureType::Texture1DBuffer: - return false; - } - return false; -} - -[[nodiscard]] constexpr bool HasTwoDimsPerLayer(ImageType type) { - switch (type) { - case ImageType::e2D: - case ImageType::e3D: - case ImageType::Linear: - return true; - case ImageType::e1D: - case ImageType::Buffer: - return false; - } - UNREACHABLE_MSG("Invalid image type={}", static_cast<int>(type)); -} - [[nodiscard]] constexpr std::pair<int, int> Samples(int num_samples) { switch (num_samples) { case 1: @@ -279,7 +245,7 @@ template <u32 GOB_EXTENT> const bool is_small = IsSmallerThanGobSize(blocks, gob, info.block.depth); const u32 alignment = is_small ? 0 : info.tile_width_spacing; return Extent2D{ - .width = Common::AlignBits(gobs.width, alignment), + .width = Common::AlignUpLog2(gobs.width, alignment), .height = gobs.height, }; } @@ -352,7 +318,7 @@ template <u32 GOB_EXTENT> // https://github.com/Ryujinx/Ryujinx/blob/1c9aba6de1520aea5480c032e0ff5664ac1bb36f/Ryujinx.Graphics.Texture/SizeCalculator.cs#L134 if (tile_width_spacing > 0) { const u32 alignment_log2 = GOB_SIZE_SHIFT + tile_width_spacing + block.height + block.depth; - return Common::AlignBits(size_bytes, alignment_log2); + return Common::AlignUpLog2(size_bytes, alignment_log2); } const u32 aligned_height = Common::AlignUp(size.height, tile_size_y); while (block.height != 0 && aligned_height <= (1U << (block.height - 1)) * GOB_SIZE_Y) { @@ -528,9 +494,9 @@ template <u32 GOB_EXTENT> const u32 alignment = StrideAlignment(num_tiles, info.block, bpp_log2, info.tile_width_spacing); const Extent3D mip_block = AdjustMipBlockSize(num_tiles, info.block, 0); return Extent3D{ - .width = Common::AlignBits(num_tiles.width, alignment), - .height = Common::AlignBits(num_tiles.height, GOB_SIZE_Y_SHIFT + mip_block.height), - .depth = Common::AlignBits(num_tiles.depth, GOB_SIZE_Z_SHIFT + mip_block.depth), + .width = Common::AlignUpLog2(num_tiles.width, alignment), + .height = Common::AlignUpLog2(num_tiles.height, GOB_SIZE_Y_SHIFT + mip_block.height), + .depth = Common::AlignUpLog2(num_tiles.depth, GOB_SIZE_Z_SHIFT + mip_block.depth), }; } @@ -679,7 +645,7 @@ u32 CalculateLayerSize(const ImageInfo& info) noexcept { } std::array<u32, MAX_MIP_LEVELS> CalculateMipLevelOffsets(const ImageInfo& info) noexcept { - ASSERT(info.resources.levels <= MAX_MIP_LEVELS); + ASSERT(info.resources.levels <= static_cast<s32>(MAX_MIP_LEVELS)); const LevelInfo level_info = MakeLevelInfo(info); std::array<u32, MAX_MIP_LEVELS> offsets{}; u32 offset = 0; @@ -1069,13 +1035,13 @@ bool IsPitchLinearSameSize(const ImageInfo& lhs, const ImageInfo& rhs, bool stri std::optional<OverlapResult> ResolveOverlap(const ImageInfo& new_info, GPUVAddr gpu_addr, VAddr cpu_addr, const ImageBase& overlap, - bool strict_size, bool broken_views) { + bool strict_size, bool broken_views, bool native_bgr) { ASSERT(new_info.type != ImageType::Linear); ASSERT(overlap.info.type != ImageType::Linear); if (!IsLayerStrideCompatible(new_info, overlap.info)) { return std::nullopt; } - if (!IsViewCompatible(overlap.info.format, new_info.format, broken_views)) { + if (!IsViewCompatible(overlap.info.format, new_info.format, broken_views, native_bgr)) { return std::nullopt; } if (gpu_addr == overlap.gpu_addr) { @@ -1119,14 +1085,14 @@ bool IsLayerStrideCompatible(const ImageInfo& lhs, const ImageInfo& rhs) { std::optional<SubresourceBase> FindSubresource(const ImageInfo& candidate, const ImageBase& image, GPUVAddr candidate_addr, RelaxedOptions options, - bool broken_views) { + bool broken_views, bool native_bgr) { const std::optional<SubresourceBase> base = image.TryFindBase(candidate_addr); if (!base) { return std::nullopt; } const ImageInfo& existing = image.info; if (False(options & RelaxedOptions::Format)) { - if (!IsViewCompatible(existing.format, candidate.format, broken_views)) { + if (!IsViewCompatible(existing.format, candidate.format, broken_views, native_bgr)) { return std::nullopt; } } @@ -1163,8 +1129,9 @@ std::optional<SubresourceBase> FindSubresource(const ImageInfo& candidate, const } bool IsSubresource(const ImageInfo& candidate, const ImageBase& image, GPUVAddr candidate_addr, - RelaxedOptions options, bool broken_views) { - return FindSubresource(candidate, image, candidate_addr, options, broken_views).has_value(); + RelaxedOptions options, bool broken_views, bool native_bgr) { + return FindSubresource(candidate, image, candidate_addr, options, broken_views, native_bgr) + .has_value(); } void DeduceBlitImages(ImageInfo& dst_info, ImageInfo& src_info, const ImageBase* dst, @@ -1193,25 +1160,35 @@ u32 MapSizeBytes(const ImageBase& image) { } } -using P = PixelFormat; - -static_assert(CalculateLevelSize(LevelInfo{{1920, 1080}, {0, 2, 0}, {1, 1}, 2, 0}, 0) == 0x7f8000); -static_assert(CalculateLevelSize(LevelInfo{{32, 32}, {0, 0, 4}, {1, 1}, 4, 0}, 0) == 0x4000); - -static_assert(CalculateLevelOffset(P::R8_SINT, {1920, 1080}, {0, 2}, 1, 0, 7) == 0x2afc00); -static_assert(CalculateLevelOffset(P::ASTC_2D_12X12_UNORM, {8192, 4096}, {0, 2}, 1, 0, 12) == - 0x50d200); - -static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 0) == 0); -static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 1) == 0x400000); -static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 2) == 0x500000); -static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 3) == 0x540000); -static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 4) == 0x550000); -static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 5) == 0x554000); -static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 6) == 0x555000); -static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 7) == 0x555400); -static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 8) == 0x555600); -static_assert(CalculateLevelOffset(P::A8B8G8R8_UNORM, {1024, 1024}, {0, 4}, 1, 0, 9) == 0x555800); +static_assert(CalculateLevelSize(LevelInfo{{1920, 1080, 1}, {0, 2, 0}, {1, 1}, 2, 0}, 0) == + 0x7f8000); +static_assert(CalculateLevelSize(LevelInfo{{32, 32, 1}, {0, 0, 4}, {1, 1}, 4, 0}, 0) == 0x4000); + +static_assert(CalculateLevelOffset(PixelFormat::R8_SINT, {1920, 1080, 1}, {0, 2, 0}, 1, 0, 7) == + 0x2afc00); +static_assert(CalculateLevelOffset(PixelFormat::ASTC_2D_12X12_UNORM, {8192, 4096, 1}, {0, 2, 0}, 1, + 0, 12) == 0x50d200); + +static_assert(CalculateLevelOffset(PixelFormat::A8B8G8R8_UNORM, {1024, 1024, 1}, {0, 4, 0}, 1, 0, + 0) == 0); +static_assert(CalculateLevelOffset(PixelFormat::A8B8G8R8_UNORM, {1024, 1024, 1}, {0, 4, 0}, 1, 0, + 1) == 0x400000); +static_assert(CalculateLevelOffset(PixelFormat::A8B8G8R8_UNORM, {1024, 1024, 1}, {0, 4, 0}, 1, 0, + 2) == 0x500000); +static_assert(CalculateLevelOffset(PixelFormat::A8B8G8R8_UNORM, {1024, 1024, 1}, {0, 4, 0}, 1, 0, + 3) == 0x540000); +static_assert(CalculateLevelOffset(PixelFormat::A8B8G8R8_UNORM, {1024, 1024, 1}, {0, 4, 0}, 1, 0, + 4) == 0x550000); +static_assert(CalculateLevelOffset(PixelFormat::A8B8G8R8_UNORM, {1024, 1024, 1}, {0, 4, 0}, 1, 0, + 5) == 0x554000); +static_assert(CalculateLevelOffset(PixelFormat::A8B8G8R8_UNORM, {1024, 1024, 1}, {0, 4, 0}, 1, 0, + 6) == 0x555000); +static_assert(CalculateLevelOffset(PixelFormat::A8B8G8R8_UNORM, {1024, 1024, 1}, {0, 4, 0}, 1, 0, + 7) == 0x555400); +static_assert(CalculateLevelOffset(PixelFormat::A8B8G8R8_UNORM, {1024, 1024, 1}, {0, 4, 0}, 1, 0, + 8) == 0x555600); +static_assert(CalculateLevelOffset(PixelFormat::A8B8G8R8_UNORM, {1024, 1024, 1}, {0, 4, 0}, 1, 0, + 9) == 0x555800); constexpr u32 ValidateLayerSize(PixelFormat format, u32 width, u32 height, u32 block_height, u32 tile_width_spacing, u32 level) { @@ -1221,13 +1198,14 @@ constexpr u32 ValidateLayerSize(PixelFormat format, u32 width, u32 height, u32 b return AlignLayerSize(offset, size, block, DefaultBlockHeight(format), tile_width_spacing); } -static_assert(ValidateLayerSize(P::ASTC_2D_12X12_UNORM, 8192, 4096, 2, 0, 12) == 0x50d800); -static_assert(ValidateLayerSize(P::A8B8G8R8_UNORM, 1024, 1024, 2, 0, 10) == 0x556000); -static_assert(ValidateLayerSize(P::BC3_UNORM, 128, 128, 2, 0, 8) == 0x6000); +static_assert(ValidateLayerSize(PixelFormat::ASTC_2D_12X12_UNORM, 8192, 4096, 2, 0, 12) == + 0x50d800); +static_assert(ValidateLayerSize(PixelFormat::A8B8G8R8_UNORM, 1024, 1024, 2, 0, 10) == 0x556000); +static_assert(ValidateLayerSize(PixelFormat::BC3_UNORM, 128, 128, 2, 0, 8) == 0x6000); -static_assert(ValidateLayerSize(P::A8B8G8R8_UNORM, 518, 572, 4, 3, 1) == 0x190000, +static_assert(ValidateLayerSize(PixelFormat::A8B8G8R8_UNORM, 518, 572, 4, 3, 1) == 0x190000, "Tile width spacing is not working"); -static_assert(ValidateLayerSize(P::BC5_UNORM, 1024, 1024, 3, 4, 11) == 0x160000, +static_assert(ValidateLayerSize(PixelFormat::BC5_UNORM, 1024, 1024, 3, 4, 11) == 0x160000, "Compressed tile width spacing is not working"); } // namespace VideoCommon diff --git a/src/video_core/texture_cache/util.h b/src/video_core/texture_cache/util.h index 52a9207d6..4d0072867 100644 --- a/src/video_core/texture_cache/util.h +++ b/src/video_core/texture_cache/util.h @@ -87,7 +87,8 @@ void SwizzleImage(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr, const Ima [[nodiscard]] std::optional<OverlapResult> ResolveOverlap(const ImageInfo& new_info, GPUVAddr gpu_addr, VAddr cpu_addr, const ImageBase& overlap, - bool strict_size, bool broken_views); + bool strict_size, bool broken_views, + bool native_bgr); [[nodiscard]] bool IsLayerStrideCompatible(const ImageInfo& lhs, const ImageInfo& rhs); @@ -95,11 +96,11 @@ void SwizzleImage(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr, const Ima const ImageBase& image, GPUVAddr candidate_addr, RelaxedOptions options, - bool broken_views); + bool broken_views, bool native_bgr); [[nodiscard]] bool IsSubresource(const ImageInfo& candidate, const ImageBase& image, - GPUVAddr candidate_addr, RelaxedOptions options, - bool broken_views); + GPUVAddr candidate_addr, RelaxedOptions options, bool broken_views, + bool native_bgr); void DeduceBlitImages(ImageInfo& dst_info, ImageInfo& src_info, const ImageBase* dst, const ImageBase* src); |
