diff options
Diffstat (limited to 'src/video_core')
94 files changed, 308 insertions, 232 deletions
diff --git a/src/video_core/buffer_cache/buffer_base.h b/src/video_core/buffer_cache/buffer_base.h index be2113f5a..10975884b 100644 --- a/src/video_core/buffer_cache/buffer_base.h +++ b/src/video_core/buffer_cache/buffer_base.h @@ -212,7 +212,7 @@ public: void FlushCachedWrites() noexcept { flags &= ~BufferFlagBits::CachedWrites; const u64 num_words = NumWords(); - const u64* const cached_words = Array<Type::CachedCPU>(); + u64* const cached_words = Array<Type::CachedCPU>(); u64* const untracked_words = Array<Type::Untracked>(); u64* const cpu_words = Array<Type::CPU>(); for (u64 word_index = 0; word_index < num_words; ++word_index) { @@ -220,6 +220,7 @@ public: NotifyRasterizer<false>(word_index, untracked_words[word_index], cached_bits); untracked_words[word_index] |= cached_bits; cpu_words[word_index] |= cached_bits; + cached_words[word_index] = 0; } } diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h index 6d8955ca3..21bfb76a4 100644 --- a/src/video_core/buffer_cache/buffer_cache.h +++ b/src/video_core/buffer_cache/buffer_cache.h @@ -11,7 +11,6 @@ #include <mutex> #include <numeric> #include <span> -#include <unordered_map> #include <vector> #include <boost/container/small_vector.hpp> @@ -22,7 +21,6 @@ #include "common/literals.h" #include "common/lru_cache.h" #include "common/microprofile.h" -#include "common/scope_exit.h" #include "common/settings.h" #include "core/memory.h" #include "video_core/buffer_cache/buffer_base.h" @@ -78,8 +76,9 @@ class BufferCache { static constexpr BufferId NULL_BUFFER_ID{0}; - static constexpr u64 EXPECTED_MEMORY = 512_MiB; - static constexpr u64 CRITICAL_MEMORY = 1_GiB; + static constexpr s64 DEFAULT_EXPECTED_MEMORY = 512_MiB; + static constexpr s64 DEFAULT_CRITICAL_MEMORY = 1_GiB; + static constexpr s64 TARGET_THRESHOLD = 4_GiB; using Maxwell = Tegra::Engines::Maxwell3D::Regs; @@ -438,6 +437,8 @@ private: Common::LeastRecentlyUsedCache<LRUItemParams> lru_cache; u64 frame_tick = 0; u64 total_used_memory = 0; + u64 minimum_memory = 0; + u64 critical_memory = 0; std::array<BufferId, ((1ULL << 39) >> PAGE_BITS)> page_table; }; @@ -453,11 +454,30 @@ BufferCache<P>::BufferCache(VideoCore::RasterizerInterface& rasterizer_, // Ensure the first slot is used for the null buffer void(slot_buffers.insert(runtime, NullBufferParams{})); common_ranges.clear(); + + if (!runtime.CanReportMemoryUsage()) { + minimum_memory = DEFAULT_EXPECTED_MEMORY; + critical_memory = DEFAULT_CRITICAL_MEMORY; + return; + } + + const s64 device_memory = static_cast<s64>(runtime.GetDeviceLocalMemory()); + const s64 min_spacing_expected = device_memory - 1_GiB - 512_MiB; + const s64 min_spacing_critical = device_memory - 1_GiB; + const s64 mem_threshold = std::min(device_memory, TARGET_THRESHOLD); + const s64 min_vacancy_expected = (6 * mem_threshold) / 10; + const s64 min_vacancy_critical = (3 * mem_threshold) / 10; + minimum_memory = static_cast<u64>( + std::max(std::min(device_memory - min_vacancy_expected, min_spacing_expected), + DEFAULT_EXPECTED_MEMORY)); + critical_memory = static_cast<u64>( + std::max(std::min(device_memory - min_vacancy_critical, min_spacing_critical), + DEFAULT_CRITICAL_MEMORY)); } template <class P> void BufferCache<P>::RunGarbageCollector() { - const bool aggressive_gc = total_used_memory >= CRITICAL_MEMORY; + const bool aggressive_gc = total_used_memory >= critical_memory; const u64 ticks_to_destroy = aggressive_gc ? 60 : 120; int num_iterations = aggressive_gc ? 64 : 32; const auto clean_up = [this, &num_iterations](BufferId buffer_id) { @@ -488,7 +508,11 @@ void BufferCache<P>::TickFrame() { const bool skip_preferred = hits * 256 < shots * 251; uniform_buffer_skip_cache_size = skip_preferred ? DEFAULT_SKIP_CACHE_SIZE : 0; - if (total_used_memory >= EXPECTED_MEMORY) { + // If we can obtain the memory info, use it instead of the estimate. + if (runtime.CanReportMemoryUsage()) { + total_used_memory = runtime.GetDeviceMemoryUsage(); + } + if (total_used_memory >= minimum_memory) { RunGarbageCollector(); } ++frame_tick; diff --git a/src/video_core/cdma_pusher.cpp b/src/video_core/cdma_pusher.cpp index a8c4b4415..8dd840558 100644 --- a/src/video_core/cdma_pusher.cpp +++ b/src/video_core/cdma_pusher.cpp @@ -23,11 +23,9 @@ #include "command_classes/nvdec.h" #include "command_classes/vic.h" #include "video_core/cdma_pusher.h" -#include "video_core/command_classes/nvdec_common.h" #include "video_core/command_classes/sync_manager.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/gpu.h" -#include "video_core/memory_manager.h" namespace Tegra { CDmaPusher::CDmaPusher(GPU& gpu_) diff --git a/src/video_core/cdma_pusher.h b/src/video_core/cdma_pusher.h index 87b49d6ea..cb1d16b71 100644 --- a/src/video_core/cdma_pusher.h +++ b/src/video_core/cdma_pusher.h @@ -8,6 +8,7 @@ #include <vector> #include "common/bit_field.h" +#include "common/common_funcs.h" #include "common/common_types.h" namespace Tegra { diff --git a/src/video_core/command_classes/codecs/codec.cpp b/src/video_core/command_classes/codecs/codec.cpp index c9ab11e53..40f7755e8 100644 --- a/src/video_core/command_classes/codecs/codec.cpp +++ b/src/video_core/command_classes/codecs/codec.cpp @@ -3,7 +3,6 @@ // Refer to the license.txt file included. #include <algorithm> -#include <cstdio> #include <fstream> #include <vector> #include "common/assert.h" diff --git a/src/video_core/command_classes/codecs/codec.h b/src/video_core/command_classes/codecs/codec.h index de5672155..661673b4e 100644 --- a/src/video_core/command_classes/codecs/codec.h +++ b/src/video_core/command_classes/codecs/codec.h @@ -7,7 +7,7 @@ #include <memory> #include <string_view> #include <queue> -#include "common/common_types.h" + #include "video_core/command_classes/nvdec_common.h" extern "C" { diff --git a/src/video_core/command_classes/codecs/vp8.cpp b/src/video_core/command_classes/codecs/vp8.cpp index 32ad0ec16..2f280cb7c 100644 --- a/src/video_core/command_classes/codecs/vp8.cpp +++ b/src/video_core/command_classes/codecs/vp8.cpp @@ -2,7 +2,6 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include <array> #include <vector> #include "video_core/command_classes/codecs/vp8.h" diff --git a/src/video_core/command_classes/codecs/vp9_types.h b/src/video_core/command_classes/codecs/vp9_types.h index 3b1ed4b3a..af1290016 100644 --- a/src/video_core/command_classes/codecs/vp9_types.h +++ b/src/video_core/command_classes/codecs/vp9_types.h @@ -5,7 +5,6 @@ #pragma once #include <array> -#include <cstring> #include <vector> #include "common/common_funcs.h" #include "common/common_types.h" diff --git a/src/video_core/command_classes/host1x.h b/src/video_core/command_classes/host1x.h index 7e94799dd..736d2fd0c 100644 --- a/src/video_core/command_classes/host1x.h +++ b/src/video_core/command_classes/host1x.h @@ -4,8 +4,6 @@ #pragma once -#include <vector> -#include "common/common_funcs.h" #include "common/common_types.h" namespace Tegra { diff --git a/src/video_core/dma_pusher.cpp b/src/video_core/dma_pusher.cpp index 8d28bd884..0b51e402a 100644 --- a/src/video_core/dma_pusher.cpp +++ b/src/video_core/dma_pusher.cpp @@ -6,7 +6,6 @@ #include "common/microprofile.h" #include "common/settings.h" #include "core/core.h" -#include "core/memory.h" #include "video_core/dma_pusher.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/gpu.h" diff --git a/src/video_core/engines/fermi_2d.h b/src/video_core/engines/fermi_2d.h index d76c5ed56..4c0568c4c 100644 --- a/src/video_core/engines/fermi_2d.h +++ b/src/video_core/engines/fermi_2d.h @@ -9,7 +9,6 @@ #include "common/bit_field.h" #include "common/common_funcs.h" #include "common/common_types.h" -#include "common/math_util.h" #include "video_core/engines/engine_interface.h" #include "video_core/gpu.h" diff --git a/src/video_core/engines/kepler_compute.cpp b/src/video_core/engines/kepler_compute.cpp index 5a1c12076..f7ff92c57 100644 --- a/src/video_core/engines/kepler_compute.cpp +++ b/src/video_core/engines/kepler_compute.cpp @@ -10,7 +10,6 @@ #include "video_core/engines/maxwell_3d.h" #include "video_core/memory_manager.h" #include "video_core/rasterizer_interface.h" -#include "video_core/renderer_base.h" #include "video_core/textures/decoders.h" namespace Tegra::Engines { diff --git a/src/video_core/engines/kepler_compute.h b/src/video_core/engines/kepler_compute.h index f8b8d06ac..c6b8adb56 100644 --- a/src/video_core/engines/kepler_compute.h +++ b/src/video_core/engines/kepler_compute.h @@ -12,7 +12,6 @@ #include "common/common_types.h" #include "video_core/engines/engine_interface.h" #include "video_core/engines/engine_upload.h" -#include "video_core/gpu.h" #include "video_core/textures/texture.h" namespace Core { diff --git a/src/video_core/engines/kepler_memory.cpp b/src/video_core/engines/kepler_memory.cpp index 8aed16caa..f930e02b6 100644 --- a/src/video_core/engines/kepler_memory.cpp +++ b/src/video_core/engines/kepler_memory.cpp @@ -9,8 +9,6 @@ #include "video_core/engines/maxwell_3d.h" #include "video_core/memory_manager.h" #include "video_core/rasterizer_interface.h" -#include "video_core/renderer_base.h" -#include "video_core/textures/decoders.h" namespace Tegra::Engines { diff --git a/src/video_core/engines/kepler_memory.h b/src/video_core/engines/kepler_memory.h index 949e2fae1..4cb4a3d2d 100644 --- a/src/video_core/engines/kepler_memory.h +++ b/src/video_core/engines/kepler_memory.h @@ -6,13 +6,11 @@ #include <array> #include <cstddef> -#include <vector> #include "common/bit_field.h" #include "common/common_funcs.h" #include "common/common_types.h" #include "video_core/engines/engine_interface.h" #include "video_core/engines/engine_upload.h" -#include "video_core/gpu.h" namespace Core { class System; diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h index 357a74c70..3f5b38e55 100644 --- a/src/video_core/engines/maxwell_3d.h +++ b/src/video_core/engines/maxwell_3d.h @@ -10,7 +10,6 @@ #include <limits> #include <optional> #include <type_traits> -#include <unordered_map> #include <vector> #include "common/assert.h" diff --git a/src/video_core/engines/maxwell_dma.h b/src/video_core/engines/maxwell_dma.h index 2692cac8a..9d0c77793 100644 --- a/src/video_core/engines/maxwell_dma.h +++ b/src/video_core/engines/maxwell_dma.h @@ -8,10 +8,8 @@ #include <cstddef> #include <vector> #include "common/bit_field.h" -#include "common/common_funcs.h" #include "common/common_types.h" #include "video_core/engines/engine_interface.h" -#include "video_core/gpu.h" namespace Core { class System; diff --git a/src/video_core/framebuffer_config.h b/src/video_core/framebuffer_config.h index b1d455e30..93349bb78 100644 --- a/src/video_core/framebuffer_config.h +++ b/src/video_core/framebuffer_config.h @@ -6,41 +6,22 @@ #include "common/common_types.h" #include "common/math_util.h" +#include "core/hle/service/nvflinger/buffer_transform_flags.h" +#include "core/hle/service/nvflinger/pixel_format.h" namespace Tegra { + /** * Struct describing framebuffer configuration */ struct FramebufferConfig { - enum class PixelFormat : u32 { - A8B8G8R8_UNORM = 1, - RGB565_UNORM = 4, - B8G8R8A8_UNORM = 5, - }; - - enum class TransformFlags : u32 { - /// No transform flags are set - Unset = 0x00, - /// Flip source image horizontally (around the vertical axis) - FlipH = 0x01, - /// Flip source image vertically (around the horizontal axis) - FlipV = 0x02, - /// Rotate source image 90 degrees clockwise - Rotate90 = 0x04, - /// Rotate source image 180 degrees - Rotate180 = 0x03, - /// Rotate source image 270 degrees clockwise - Rotate270 = 0x07, - }; - VAddr address{}; u32 offset{}; u32 width{}; u32 height{}; u32 stride{}; - PixelFormat pixel_format{}; - - TransformFlags transform_flags{}; + Service::android::PixelFormat pixel_format{}; + Service::android::BufferTransformFlags transform_flags{}; Common::Rectangle<int> crop_rect; }; diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h index 26b8ea233..97c029140 100644 --- a/src/video_core/gpu.h +++ b/src/video_core/gpu.h @@ -8,6 +8,7 @@ #include "common/bit_field.h" #include "common/common_types.h" +#include "core/hle/service/nvdrv/nvdata.h" #include "video_core/cdma_pusher.h" #include "video_core/framebuffer_config.h" diff --git a/src/video_core/memory_manager.cpp b/src/video_core/memory_manager.cpp index 4ff3fa268..722ebd9ad 100644 --- a/src/video_core/memory_manager.cpp +++ b/src/video_core/memory_manager.cpp @@ -11,7 +11,6 @@ #include "core/hle/kernel/k_page_table.h" #include "core/hle/kernel/k_process.h" #include "core/memory.h" -#include "video_core/gpu.h" #include "video_core/memory_manager.h" #include "video_core/rasterizer_interface.h" #include "video_core/renderer_base.h" diff --git a/src/video_core/query_cache.h b/src/video_core/query_cache.h index 0173b54d8..8a84bcfa9 100644 --- a/src/video_core/query_cache.h +++ b/src/video_core/query_cache.h @@ -19,7 +19,6 @@ #include "common/assert.h" #include "common/settings.h" #include "video_core/engines/maxwell_3d.h" -#include "video_core/gpu.h" #include "video_core/memory_manager.h" #include "video_core/rasterizer_interface.h" diff --git a/src/video_core/renderer_base.cpp b/src/video_core/renderer_base.cpp index a99c33c37..c9627cffd 100644 --- a/src/video_core/renderer_base.cpp +++ b/src/video_core/renderer_base.cpp @@ -3,7 +3,6 @@ // Refer to the license.txt file included. #include "common/logging/log.h" -#include "common/settings.h" #include "core/frontend/emu_window.h" #include "video_core/renderer_base.h" diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.cpp b/src/video_core/renderer_opengl/gl_buffer_cache.cpp index d4dd10bb6..f1f7b384b 100644 --- a/src/video_core/renderer_opengl/gl_buffer_cache.cpp +++ b/src/video_core/renderer_opengl/gl_buffer_cache.cpp @@ -135,6 +135,20 @@ BufferCacheRuntime::BufferCacheRuntime(const Device& device_) buffer.Create(); glNamedBufferData(buffer.handle, 0x10'000, nullptr, GL_STREAM_COPY); } + + device_access_memory = [this]() -> u64 { + if (device.CanReportMemoryUsage()) { + return device.GetCurrentDedicatedVideoMemory() + 512_MiB; + } + return 2_GiB; // Return minimum requirements + }(); +} + +u64 BufferCacheRuntime::GetDeviceMemoryUsage() const { + if (device.CanReportMemoryUsage()) { + return device_access_memory - device.GetCurrentDedicatedVideoMemory(); + } + return 2_GiB; } void BufferCacheRuntime::CopyBuffer(Buffer& dst_buffer, Buffer& src_buffer, diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.h b/src/video_core/renderer_opengl/gl_buffer_cache.h index 060d36427..a8699f28c 100644 --- a/src/video_core/renderer_opengl/gl_buffer_cache.h +++ b/src/video_core/renderer_opengl/gl_buffer_cache.h @@ -7,9 +7,7 @@ #include <array> #include <span> -#include "common/alignment.h" #include "common/common_types.h" -#include "common/dynamic_library.h" #include "video_core/buffer_cache/buffer_cache.h" #include "video_core/rasterizer_interface.h" #include "video_core/renderer_opengl/gl_device.h" @@ -91,6 +89,8 @@ public: void BindImageBuffer(Buffer& buffer, u32 offset, u32 size, VideoCore::Surface::PixelFormat format); + u64 GetDeviceMemoryUsage() const; + void BindFastUniformBuffer(size_t stage, u32 binding_index, u32 size) { const GLuint handle = fast_uniforms[stage][binding_index].handle; const GLsizeiptr gl_size = static_cast<GLsizeiptr>(size); @@ -153,6 +153,14 @@ public: use_storage_buffers = use_storage_buffers_; } + u64 GetDeviceLocalMemory() const { + return device_access_memory; + } + + bool CanReportMemoryUsage() const { + return device.CanReportMemoryUsage(); + } + private: static constexpr std::array PABO_LUT{ GL_VERTEX_PROGRAM_PARAMETER_BUFFER_NV, GL_TESS_CONTROL_PROGRAM_PARAMETER_BUFFER_NV, @@ -186,6 +194,8 @@ private: std::array<OGLBuffer, VideoCommon::NUM_COMPUTE_UNIFORM_BUFFERS> copy_compute_uniforms; u32 index_buffer_offset = 0; + + u64 device_access_memory; }; struct BufferCacheParams { diff --git a/src/video_core/renderer_opengl/gl_compute_pipeline.h b/src/video_core/renderer_opengl/gl_compute_pipeline.h index 50c676365..b0d183b46 100644 --- a/src/video_core/renderer_opengl/gl_compute_pipeline.h +++ b/src/video_core/renderer_opengl/gl_compute_pipeline.h @@ -6,7 +6,6 @@ #include <array> #include <type_traits> -#include <utility> #include "common/common_types.h" #include "shader_recompiler/shader_info.h" diff --git a/src/video_core/renderer_opengl/gl_device.cpp b/src/video_core/renderer_opengl/gl_device.cpp index e62912a22..656dd7eb0 100644 --- a/src/video_core/renderer_opengl/gl_device.cpp +++ b/src/video_core/renderer_opengl/gl_device.cpp @@ -6,8 +6,6 @@ #include <array> #include <cstddef> #include <cstdlib> -#include <cstring> -#include <limits> #include <optional> #include <span> #include <stdexcept> @@ -15,13 +13,15 @@ #include <glad/glad.h> +#include "common/literals.h" #include "common/logging/log.h" -#include "common/scope_exit.h" #include "common/settings.h" #include "shader_recompiler/stage.h" #include "video_core/renderer_opengl/gl_device.h" #include "video_core/renderer_opengl/gl_resource_manager.h" +using namespace Common::Literals; + namespace OpenGL { namespace { constexpr std::array LIMIT_UBOS = { @@ -168,6 +168,7 @@ Device::Device() { has_sparse_texture_2 = GLAD_GL_ARB_sparse_texture2; warp_size_potentially_larger_than_guest = !is_nvidia && !is_intel; need_fastmath_off = is_nvidia; + can_report_memory = GLAD_GL_NVX_gpu_memory_info; // At the moment of writing this, only Nvidia's driver optimizes BufferSubData on exclusive // uniform buffers as "push constants" @@ -279,4 +280,10 @@ void main() { })"); } +u64 Device::GetCurrentDedicatedVideoMemory() const { + GLint cur_avail_mem_kb = 0; + glGetIntegerv(GL_GPU_MEMORY_INFO_DEDICATED_VIDMEM_NVX, &cur_avail_mem_kb); + return static_cast<u64>(cur_avail_mem_kb) * 1_KiB; +} + } // namespace OpenGL diff --git a/src/video_core/renderer_opengl/gl_device.h b/src/video_core/renderer_opengl/gl_device.h index 95c2e8d38..9bb0b9148 100644 --- a/src/video_core/renderer_opengl/gl_device.h +++ b/src/video_core/renderer_opengl/gl_device.h @@ -20,6 +20,8 @@ public: [[nodiscard]] std::string GetVendorName() const; + u64 GetCurrentDedicatedVideoMemory() const; + u32 GetMaxUniformBuffers(Shader::Stage stage) const noexcept { return max_uniform_buffers[static_cast<size_t>(stage)]; } @@ -168,6 +170,10 @@ public: return vendor_name == "ATI Technologies Inc."; } + bool CanReportMemoryUsage() const { + return can_report_memory; + } + private: static bool TestVariableAoffi(); static bool TestPreciseBug(); @@ -210,6 +216,7 @@ private: bool need_fastmath_off{}; bool has_cbuf_ftou_bug{}; bool has_bool_ref_bug{}; + bool can_report_memory{}; std::string vendor_name; }; diff --git a/src/video_core/renderer_opengl/gl_graphics_pipeline.h b/src/video_core/renderer_opengl/gl_graphics_pipeline.h index 311d49f3f..4f8049717 100644 --- a/src/video_core/renderer_opengl/gl_graphics_pipeline.h +++ b/src/video_core/renderer_opengl/gl_graphics_pipeline.h @@ -14,7 +14,6 @@ #include "common/common_types.h" #include "shader_recompiler/shader_info.h" #include "video_core/engines/maxwell_3d.h" -#include "video_core/memory_manager.h" #include "video_core/renderer_opengl/gl_buffer_cache.h" #include "video_core/renderer_opengl/gl_resource_manager.h" #include "video_core/renderer_opengl/gl_texture_cache.h" diff --git a/src/video_core/renderer_opengl/gl_query_cache.cpp b/src/video_core/renderer_opengl/gl_query_cache.cpp index acebbf5f4..9e6c50055 100644 --- a/src/video_core/renderer_opengl/gl_query_cache.cpp +++ b/src/video_core/renderer_opengl/gl_query_cache.cpp @@ -3,15 +3,12 @@ // Refer to the license.txt file included. #include <algorithm> -#include <cstring> #include <memory> -#include <unordered_map> #include <utility> #include <vector> #include <glad/glad.h> -#include "common/assert.h" #include "core/core.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/memory_manager.h" diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 142412a8e..e6f9ece8b 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -6,9 +6,7 @@ #include <array> #include <bitset> #include <memory> -#include <string> #include <string_view> -#include <tuple> #include <utility> #include <glad/glad.h> @@ -17,8 +15,9 @@ #include "common/logging/log.h" #include "common/math_util.h" #include "common/microprofile.h" +#include "common/scope_exit.h" #include "common/settings.h" -#include "core/memory.h" + #include "video_core/engines/kepler_compute.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/memory_manager.h" @@ -212,6 +211,7 @@ void RasterizerOpenGL::Clear() { void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) { MICROPROFILE_SCOPE(OpenGL_Drawing); + SCOPE_EXIT({ gpu.TickWork(); }); query_cache.UpdateCounters(); GraphicsPipeline* const pipeline{shader_cache.CurrentGraphicsPipeline()}; @@ -267,8 +267,6 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) { ++num_queued_commands; has_written_global_memory |= pipeline->WritesGlobalMemory(); - - gpu.TickWork(); } void RasterizerOpenGL::DispatchCompute() { diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h index 98f6fd342..c79461d59 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.h +++ b/src/video_core/renderer_opengl/gl_rasterizer.h @@ -5,20 +5,14 @@ #pragma once #include <array> -#include <atomic> #include <cstddef> -#include <memory> #include <optional> -#include <tuple> -#include <utility> #include <boost/container/static_vector.hpp> #include <glad/glad.h> #include "common/common_types.h" -#include "video_core/engines/const_buffer_info.h" -#include "video_core/engines/maxwell_3d.h" #include "video_core/engines/maxwell_dma.h" #include "video_core/rasterizer_accelerated.h" #include "video_core/rasterizer_interface.h" @@ -26,12 +20,8 @@ #include "video_core/renderer_opengl/gl_device.h" #include "video_core/renderer_opengl/gl_fence_manager.h" #include "video_core/renderer_opengl/gl_query_cache.h" -#include "video_core/renderer_opengl/gl_resource_manager.h" #include "video_core/renderer_opengl/gl_shader_cache.h" -#include "video_core/renderer_opengl/gl_shader_manager.h" -#include "video_core/renderer_opengl/gl_state_tracker.h" #include "video_core/renderer_opengl/gl_texture_cache.h" -#include "video_core/textures/texture.h" namespace Core::Memory { class Memory; diff --git a/src/video_core/renderer_opengl/gl_resource_manager.cpp b/src/video_core/renderer_opengl/gl_resource_manager.cpp index 5e7101d28..f6839a657 100644 --- a/src/video_core/renderer_opengl/gl_resource_manager.cpp +++ b/src/video_core/renderer_opengl/gl_resource_manager.cpp @@ -3,9 +3,7 @@ // Refer to the license.txt file included. #include <string_view> -#include <utility> #include <glad/glad.h> -#include "common/common_types.h" #include "common/microprofile.h" #include "video_core/renderer_opengl/gl_resource_manager.h" #include "video_core/renderer_opengl/gl_shader_util.h" diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp index f71e01a34..6423992c3 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp @@ -14,10 +14,8 @@ #include "common/fs/fs.h" #include "common/fs/path_util.h" #include "common/logging/log.h" -#include "common/scope_exit.h" #include "common/settings.h" #include "common/thread_worker.h" -#include "core/core.h" #include "shader_recompiler/backend/glasm/emit_glasm.h" #include "shader_recompiler/backend/glsl/emit_glsl.h" #include "shader_recompiler/backend/spirv/emit_spirv.h" @@ -29,7 +27,6 @@ #include "video_core/engines/maxwell_3d.h" #include "video_core/memory_manager.h" #include "video_core/renderer_opengl/gl_rasterizer.h" -#include "video_core/renderer_opengl/gl_resource_manager.h" #include "video_core/renderer_opengl/gl_shader_cache.h" #include "video_core/renderer_opengl/gl_shader_util.h" #include "video_core/renderer_opengl/gl_state_tracker.h" diff --git a/src/video_core/renderer_opengl/gl_shader_cache.h b/src/video_core/renderer_opengl/gl_shader_cache.h index a34110b37..06d4b38bb 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.h +++ b/src/video_core/renderer_opengl/gl_shader_cache.h @@ -4,18 +4,13 @@ #pragma once -#include <array> #include <filesystem> #include <stop_token> #include <unordered_map> -#include <glad/glad.h> - #include "common/common_types.h" #include "common/thread_worker.h" -#include "shader_recompiler/frontend/ir/value.h" #include "shader_recompiler/host_translate_info.h" -#include "shader_recompiler/object_pool.h" #include "shader_recompiler/profile.h" #include "video_core/renderer_opengl/gl_compute_pipeline.h" #include "video_core/renderer_opengl/gl_graphics_pipeline.h" diff --git a/src/video_core/renderer_opengl/gl_shader_util.cpp b/src/video_core/renderer_opengl/gl_shader_util.cpp index d432072ad..129966e72 100644 --- a/src/video_core/renderer_opengl/gl_shader_util.cpp +++ b/src/video_core/renderer_opengl/gl_shader_util.cpp @@ -6,7 +6,6 @@ #include <vector> #include <glad/glad.h> -#include "common/assert.h" #include "common/logging/log.h" #include "common/settings.h" #include "video_core/renderer_opengl/gl_shader_util.h" diff --git a/src/video_core/renderer_opengl/gl_shader_util.h b/src/video_core/renderer_opengl/gl_shader_util.h index 4e1a2a8e1..a64ef37dc 100644 --- a/src/video_core/renderer_opengl/gl_shader_util.h +++ b/src/video_core/renderer_opengl/gl_shader_util.h @@ -5,14 +5,10 @@ #pragma once #include <span> -#include <string> #include <string_view> -#include <vector> #include <glad/glad.h> -#include "common/assert.h" -#include "common/logging/log.h" #include "video_core/renderer_opengl/gl_resource_manager.h" namespace OpenGL { diff --git a/src/video_core/renderer_opengl/gl_stream_buffer.h b/src/video_core/renderer_opengl/gl_stream_buffer.h index 2e67922a6..f0cb29dca 100644 --- a/src/video_core/renderer_opengl/gl_stream_buffer.h +++ b/src/video_core/renderer_opengl/gl_stream_buffer.h @@ -5,7 +5,6 @@ #pragma once #include <array> -#include <memory> #include <span> #include <utility> diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp index 3c1f79a27..8f9a65beb 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.cpp +++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp @@ -484,6 +484,13 @@ TextureCacheRuntime::TextureCacheRuntime(const Device& device_, ProgramManager& rescale_read_fbos[i].Create(); } } + + device_access_memory = [this]() -> u64 { + if (device.CanReportMemoryUsage()) { + return device.GetCurrentDedicatedVideoMemory() + 512_MiB; + } + return 2_GiB; // Return minimum requirements + }(); } TextureCacheRuntime::~TextureCacheRuntime() = default; @@ -500,13 +507,11 @@ ImageBufferMap TextureCacheRuntime::DownloadStagingBuffer(size_t size) { return download_buffers.RequestMap(size, false); } -u64 TextureCacheRuntime::GetDeviceLocalMemory() const { - if (GLAD_GL_NVX_gpu_memory_info) { - GLint cur_avail_mem_kb = 0; - glGetIntegerv(GL_GPU_MEMORY_INFO_CURRENT_AVAILABLE_VIDMEM_NVX, &cur_avail_mem_kb); - return static_cast<u64>(cur_avail_mem_kb) * 1_KiB; +u64 TextureCacheRuntime::GetDeviceMemoryUsage() const { + if (device.CanReportMemoryUsage()) { + return device_access_memory - device.GetCurrentDedicatedVideoMemory(); } - return 2_GiB; // Return minimum requirements + return 2_GiB; } void TextureCacheRuntime::CopyImage(Image& dst_image, Image& src_image, @@ -686,6 +691,7 @@ Image::Image(TextureCacheRuntime& runtime_, const VideoCommon::ImageInfo& info_, } if (IsConverted(runtime->device, info.format, info.type)) { flags |= ImageFlagBits::Converted; + flags |= ImageFlagBits::CostlyLoad; gl_internal_format = IsPixelFormatSRGB(info.format) ? GL_SRGB8_ALPHA8 : GL_RGBA8; gl_format = GL_RGBA; gl_type = GL_UNSIGNED_INT_8_8_8_8_REV; diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h index 7f425631f..53088b66e 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.h +++ b/src/video_core/renderer_opengl/gl_texture_cache.h @@ -10,6 +10,7 @@ #include <glad/glad.h> #include "shader_recompiler/shader_info.h" +#include "video_core/renderer_opengl/gl_device.h" #include "video_core/renderer_opengl/gl_resource_manager.h" #include "video_core/renderer_opengl/util_shaders.h" #include "video_core/texture_cache/image_view_base.h" @@ -21,7 +22,6 @@ struct ResolutionScalingInfo; namespace OpenGL { -class Device; class ProgramManager; class StateTracker; @@ -83,7 +83,15 @@ public: ImageBufferMap DownloadStagingBuffer(size_t size); - u64 GetDeviceLocalMemory() const; + u64 GetDeviceLocalMemory() const { + return device_access_memory; + } + + u64 GetDeviceMemoryUsage() const; + + bool CanReportMemoryUsage() const { + return device.CanReportMemoryUsage(); + } bool ShouldReinterpret([[maybe_unused]] Image& dst, [[maybe_unused]] Image& src) { return true; @@ -172,6 +180,7 @@ private: std::array<OGLFramebuffer, 4> rescale_draw_fbos; std::array<OGLFramebuffer, 4> rescale_read_fbos; const Settings::ResolutionScalingInfo& resolution; + u64 device_access_memory; }; class Image : public VideoCommon::ImageBase { diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp index f81c1b233..f8f29013a 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.cpp +++ b/src/video_core/renderer_opengl/renderer_opengl.cpp @@ -5,7 +5,6 @@ #include <algorithm> #include <cstddef> #include <cstdlib> -#include <cstring> #include <memory> #include <glad/glad.h> @@ -15,11 +14,9 @@ #include "common/microprofile.h" #include "common/settings.h" #include "common/telemetry.h" -#include "core/core.h" #include "core/core_timing.h" #include "core/frontend/emu_window.h" #include "core/memory.h" -#include "core/perf_stats.h" #include "core/telemetry_session.h" #include "video_core/host_shaders/fxaa_frag.h" #include "video_core/host_shaders/fxaa_vert.h" @@ -326,12 +323,12 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture, GLint internal_format; switch (framebuffer.pixel_format) { - case Tegra::FramebufferConfig::PixelFormat::A8B8G8R8_UNORM: + case Service::android::PixelFormat::Rgba8888: internal_format = GL_RGBA8; texture.gl_format = GL_RGBA; texture.gl_type = GL_UNSIGNED_INT_8_8_8_8_REV; break; - case Tegra::FramebufferConfig::PixelFormat::RGB565_UNORM: + case Service::android::PixelFormat::Rgb565: internal_format = GL_RGB565; texture.gl_format = GL_RGB; texture.gl_type = GL_UNSIGNED_SHORT_5_6_5; @@ -467,8 +464,8 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) { const auto& texcoords = screen_info.display_texcoords; auto left = texcoords.left; auto right = texcoords.right; - if (framebuffer_transform_flags != Tegra::FramebufferConfig::TransformFlags::Unset) { - if (framebuffer_transform_flags == Tegra::FramebufferConfig::TransformFlags::FlipV) { + if (framebuffer_transform_flags != Service::android::BufferTransformFlags::Unset) { + if (framebuffer_transform_flags == Service::android::BufferTransformFlags::FlipV) { // Flip the framebuffer vertically left = texcoords.right; right = texcoords.left; diff --git a/src/video_core/renderer_opengl/renderer_opengl.h b/src/video_core/renderer_opengl/renderer_opengl.h index cda333cad..aa206878b 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.h +++ b/src/video_core/renderer_opengl/renderer_opengl.h @@ -8,10 +8,12 @@ #include <glad/glad.h> #include "common/common_types.h" #include "common/math_util.h" + #include "video_core/renderer_base.h" #include "video_core/renderer_opengl/gl_device.h" #include "video_core/renderer_opengl/gl_rasterizer.h" #include "video_core/renderer_opengl/gl_resource_manager.h" +#include "video_core/renderer_opengl/gl_shader_manager.h" #include "video_core/renderer_opengl/gl_state_tracker.h" namespace Core { @@ -44,7 +46,7 @@ struct TextureInfo { GLsizei height; GLenum gl_format; GLenum gl_type; - Tegra::FramebufferConfig::PixelFormat pixel_format; + Service::android::PixelFormat pixel_format; }; /// Structure used for storing information about the display target for the Switch screen @@ -133,7 +135,7 @@ private: std::vector<u8> gl_framebuffer_data; /// Used for transforming the framebuffer orientation - Tegra::FramebufferConfig::TransformFlags framebuffer_transform_flags{}; + Service::android::BufferTransformFlags framebuffer_transform_flags{}; Common::Rectangle<int> framebuffer_crop_rect; }; diff --git a/src/video_core/renderer_vulkan/blit_image.cpp b/src/video_core/renderer_vulkan/blit_image.cpp index ec03cca38..abda1c490 100644 --- a/src/video_core/renderer_vulkan/blit_image.cpp +++ b/src/video_core/renderer_vulkan/blit_image.cpp @@ -367,17 +367,14 @@ BlitImageHelper::BlitImageHelper(const Device& device_, VKScheduler& scheduler_, PipelineLayoutCreateInfo(two_textures_set_layout.address()))), full_screen_vert(BuildShader(device, FULL_SCREEN_TRIANGLE_VERT_SPV)), blit_color_to_color_frag(BuildShader(device, VULKAN_BLIT_COLOR_FLOAT_FRAG_SPV)), + blit_depth_stencil_frag(BuildShader(device, VULKAN_BLIT_DEPTH_STENCIL_FRAG_SPV)), convert_depth_to_float_frag(BuildShader(device, CONVERT_DEPTH_TO_FLOAT_FRAG_SPV)), convert_float_to_depth_frag(BuildShader(device, CONVERT_FLOAT_TO_DEPTH_FRAG_SPV)), convert_abgr8_to_d24s8_frag(BuildShader(device, CONVERT_ABGR8_TO_D24S8_FRAG_SPV)), convert_d24s8_to_abgr8_frag(BuildShader(device, CONVERT_D24S8_TO_ABGR8_FRAG_SPV)), convert_s8d24_to_abgr8_frag(BuildShader(device, CONVERT_S8D24_TO_ABGR8_FRAG_SPV)), linear_sampler(device.GetLogical().CreateSampler(SAMPLER_CREATE_INFO<VK_FILTER_LINEAR>)), - nearest_sampler(device.GetLogical().CreateSampler(SAMPLER_CREATE_INFO<VK_FILTER_NEAREST>)) { - if (device.IsExtShaderStencilExportSupported()) { - blit_depth_stencil_frag = BuildShader(device, VULKAN_BLIT_DEPTH_STENCIL_FRAG_SPV); - } -} + nearest_sampler(device.GetLogical().CreateSampler(SAMPLER_CREATE_INFO<VK_FILTER_NEAREST>)) {} BlitImageHelper::~BlitImageHelper() = default; diff --git a/src/video_core/renderer_vulkan/blit_image.h b/src/video_core/renderer_vulkan/blit_image.h index 1a3944179..29ee0f67a 100644 --- a/src/video_core/renderer_vulkan/blit_image.h +++ b/src/video_core/renderer_vulkan/blit_image.h @@ -4,8 +4,6 @@ #pragma once -#include <compare> - #include "video_core/engines/fermi_2d.h" #include "video_core/renderer_vulkan/vk_descriptor_pool.h" #include "video_core/texture_cache/types.h" diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp index d70153df3..c2259ac5f 100644 --- a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp +++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp @@ -4,9 +4,6 @@ #include <algorithm> #include <cstring> -#include <tuple> - -#include <boost/functional/hash.hpp> #include "common/bit_cast.h" #include "common/cityhash.h" diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.h b/src/video_core/renderer_vulkan/maxwell_to_vk.h index 8a9616039..1c1f420f2 100644 --- a/src/video_core/renderer_vulkan/maxwell_to_vk.h +++ b/src/video_core/renderer_vulkan/maxwell_to_vk.h @@ -4,7 +4,6 @@ #pragma once -#include "common/common_types.h" #include "shader_recompiler/stage.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/surface.h" diff --git a/src/video_core/renderer_vulkan/pipeline_helper.h b/src/video_core/renderer_vulkan/pipeline_helper.h index 11c160570..c25d469e6 100644 --- a/src/video_core/renderer_vulkan/pipeline_helper.h +++ b/src/video_core/renderer_vulkan/pipeline_helper.h @@ -8,7 +8,6 @@ #include <boost/container/small_vector.hpp> -#include "common/assert.h" #include "common/common_types.h" #include "shader_recompiler/backend/spirv/emit_spirv.h" #include "shader_recompiler/shader_info.h" @@ -16,7 +15,6 @@ #include "video_core/renderer_vulkan/vk_update_descriptor.h" #include "video_core/texture_cache/texture_cache.h" #include "video_core/texture_cache/types.h" -#include "video_core/textures/texture.h" #include "video_core/vulkan_common/vulkan_device.h" namespace Vulkan { diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.cpp b/src/video_core/renderer_vulkan/renderer_vulkan.cpp index 74822814d..ef57fdfa4 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.cpp +++ b/src/video_core/renderer_vulkan/renderer_vulkan.cpp @@ -13,16 +13,15 @@ #include <fmt/format.h> #include "common/logging/log.h" +#include "common/scope_exit.h" #include "common/settings.h" #include "common/telemetry.h" -#include "core/core.h" #include "core/core_timing.h" #include "core/frontend/emu_window.h" #include "core/telemetry_session.h" #include "video_core/gpu.h" #include "video_core/renderer_vulkan/renderer_vulkan.h" #include "video_core/renderer_vulkan/vk_blit_screen.h" -#include "video_core/renderer_vulkan/vk_master_semaphore.h" #include "video_core/renderer_vulkan/vk_rasterizer.h" #include "video_core/renderer_vulkan/vk_scheduler.h" #include "video_core/renderer_vulkan/vk_state_tracker.h" diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.h b/src/video_core/renderer_vulkan/renderer_vulkan.h index 6dc985109..9680108b6 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.h +++ b/src/video_core/renderer_vulkan/renderer_vulkan.h @@ -6,7 +6,6 @@ #include <memory> #include <string> -#include <vector> #include "common/dynamic_library.h" #include "video_core/renderer_base.h" diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.cpp b/src/video_core/renderer_vulkan/vk_blit_screen.cpp index 621a6a071..d893c1952 100644 --- a/src/video_core/renderer_vulkan/vk_blit_screen.cpp +++ b/src/video_core/renderer_vulkan/vk_blit_screen.cpp @@ -6,7 +6,6 @@ #include <array> #include <cstring> #include <memory> -#include <tuple> #include <vector> #include "common/assert.h" @@ -28,7 +27,6 @@ #include "video_core/renderer_vulkan/renderer_vulkan.h" #include "video_core/renderer_vulkan/vk_blit_screen.h" #include "video_core/renderer_vulkan/vk_fsr.h" -#include "video_core/renderer_vulkan/vk_master_semaphore.h" #include "video_core/renderer_vulkan/vk_scheduler.h" #include "video_core/renderer_vulkan/vk_shader_util.h" #include "video_core/renderer_vulkan/vk_swapchain.h" @@ -96,11 +94,11 @@ std::size_t GetSizeInBytes(const Tegra::FramebufferConfig& framebuffer) { VkFormat GetFormat(const Tegra::FramebufferConfig& framebuffer) { switch (framebuffer.pixel_format) { - case Tegra::FramebufferConfig::PixelFormat::A8B8G8R8_UNORM: + case Service::android::PixelFormat::Rgba8888: return VK_FORMAT_A8B8G8R8_UNORM_PACK32; - case Tegra::FramebufferConfig::PixelFormat::RGB565_UNORM: + case Service::android::PixelFormat::Rgb565: return VK_FORMAT_R5G6B5_UNORM_PACK16; - case Tegra::FramebufferConfig::PixelFormat::B8G8R8A8_UNORM: + case Service::android::PixelFormat::Bgra8888: return VK_FORMAT_B8G8R8A8_UNORM; default: UNIMPLEMENTED_MSG("Unknown framebuffer pixel format: {}", @@ -1392,9 +1390,9 @@ void VKBlitScreen::SetVertexData(BufferData& data, const Tegra::FramebufferConfi auto right = texcoords.right; switch (framebuffer_transform_flags) { - case Tegra::FramebufferConfig::TransformFlags::Unset: + case Service::android::BufferTransformFlags::Unset: break; - case Tegra::FramebufferConfig::TransformFlags::FlipV: + case Service::android::BufferTransformFlags::FlipV: // Flip the framebuffer vertically left = texcoords.right; right = texcoords.left; diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp index 5ffd93499..def838c34 100644 --- a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp @@ -141,6 +141,18 @@ StagingBufferRef BufferCacheRuntime::DownloadStagingBuffer(size_t size) { return staging_pool.Request(size, MemoryUsage::Download); } +u64 BufferCacheRuntime::GetDeviceLocalMemory() const { + return device.GetDeviceLocalMemory(); +} + +u64 BufferCacheRuntime::GetDeviceMemoryUsage() const { + return device.GetDeviceMemoryUsage(); +} + +bool BufferCacheRuntime::CanReportMemoryUsage() const { + return device.CanReportMemoryUsage(); +} + void BufferCacheRuntime::Finish() { scheduler.Finish(); } diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.h b/src/video_core/renderer_vulkan/vk_buffer_cache.h index 1ee0d8420..d7fdd18ff 100644 --- a/src/video_core/renderer_vulkan/vk_buffer_cache.h +++ b/src/video_core/renderer_vulkan/vk_buffer_cache.h @@ -65,6 +65,12 @@ public: void Finish(); + u64 GetDeviceLocalMemory() const; + + u64 GetDeviceMemoryUsage() const; + + bool CanReportMemoryUsage() const; + [[nodiscard]] StagingBufferRef UploadStagingBuffer(size_t size); [[nodiscard]] StagingBufferRef DownloadStagingBuffer(size_t size); diff --git a/src/video_core/renderer_vulkan/vk_compute_pass.cpp b/src/video_core/renderer_vulkan/vk_compute_pass.cpp index 4d73427b4..713794410 100644 --- a/src/video_core/renderer_vulkan/vk_compute_pass.cpp +++ b/src/video_core/renderer_vulkan/vk_compute_pass.cpp @@ -3,12 +3,10 @@ // Refer to the license.txt file included. #include <array> -#include <cstring> #include <memory> #include <optional> #include <utility> -#include "common/alignment.h" #include "common/assert.h" #include "common/common_types.h" #include "common/div_ceil.h" @@ -23,7 +21,6 @@ #include "video_core/renderer_vulkan/vk_update_descriptor.h" #include "video_core/texture_cache/accelerated_swizzle.h" #include "video_core/texture_cache/types.h" -#include "video_core/textures/astc.h" #include "video_core/textures/decoders.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_compute_pipeline.h b/src/video_core/renderer_vulkan/vk_compute_pipeline.h index 8c4b0a301..c64bd9a06 100644 --- a/src/video_core/renderer_vulkan/vk_compute_pipeline.h +++ b/src/video_core/renderer_vulkan/vk_compute_pipeline.h @@ -11,7 +11,6 @@ #include "common/common_types.h" #include "common/thread_worker.h" #include "shader_recompiler/shader_info.h" -#include "video_core/memory_manager.h" #include "video_core/renderer_vulkan/vk_buffer_cache.h" #include "video_core/renderer_vulkan/vk_descriptor_pool.h" #include "video_core/renderer_vulkan/vk_texture_cache.h" diff --git a/src/video_core/renderer_vulkan/vk_fence_manager.cpp b/src/video_core/renderer_vulkan/vk_fence_manager.cpp index 3bec48d14..0c1098c8f 100644 --- a/src/video_core/renderer_vulkan/vk_fence_manager.cpp +++ b/src/video_core/renderer_vulkan/vk_fence_manager.cpp @@ -9,7 +9,6 @@ #include "video_core/renderer_vulkan/vk_scheduler.h" #include "video_core/renderer_vulkan/vk_texture_cache.h" #include "video_core/vulkan_common/vulkan_device.h" -#include "video_core/vulkan_common/vulkan_wrapper.h" namespace Vulkan { diff --git a/src/video_core/renderer_vulkan/vk_fence_manager.h b/src/video_core/renderer_vulkan/vk_fence_manager.h index 2f8322d29..cf9f4adbf 100644 --- a/src/video_core/renderer_vulkan/vk_fence_manager.h +++ b/src/video_core/renderer_vulkan/vk_fence_manager.h @@ -9,7 +9,6 @@ #include "video_core/fence_manager.h" #include "video_core/renderer_vulkan/vk_buffer_cache.h" #include "video_core/renderer_vulkan/vk_texture_cache.h" -#include "video_core/vulkan_common/vulkan_wrapper.h" namespace Core { class System; diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp index a633b73e5..27e59df73 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp @@ -16,13 +16,11 @@ #include "common/microprofile.h" #include "common/thread_worker.h" #include "core/core.h" -#include "core/memory.h" #include "shader_recompiler/backend/spirv/emit_spirv.h" #include "shader_recompiler/environment.h" #include "shader_recompiler/frontend/maxwell/control_flow.h" #include "shader_recompiler/frontend/maxwell/translate_program.h" #include "shader_recompiler/program_header.h" -#include "video_core/dirty_flags.h" #include "video_core/engines/kepler_compute.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/memory_manager.h" diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.h b/src/video_core/renderer_vulkan/vk_pipeline_cache.h index 4c135b5dd..579e25a4a 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.h +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.h @@ -7,11 +7,9 @@ #include <array> #include <cstddef> #include <filesystem> -#include <iosfwd> #include <memory> #include <type_traits> #include <unordered_map> -#include <utility> #include <vector> #include "common/common_types.h" @@ -29,7 +27,6 @@ #include "video_core/renderer_vulkan/vk_graphics_pipeline.h" #include "video_core/renderer_vulkan/vk_texture_cache.h" #include "video_core/shader_cache.h" -#include "video_core/vulkan_common/vulkan_wrapper.h" namespace Core { class System; diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index 2227d9197..fa87d37f8 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -6,15 +6,12 @@ #include <array> #include <memory> #include <mutex> -#include <vector> -#include "common/alignment.h" #include "common/assert.h" #include "common/logging/log.h" #include "common/microprofile.h" #include "common/scope_exit.h" #include "common/settings.h" -#include "core/core.h" #include "video_core/engines/kepler_compute.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/renderer_vulkan/blit_image.h" diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.h b/src/video_core/renderer_vulkan/vk_rasterizer.h index 5af2e275b..c25036fb3 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.h +++ b/src/video_core/renderer_vulkan/vk_rasterizer.h @@ -5,10 +5,6 @@ #pragma once #include <array> -#include <bitset> -#include <memory> -#include <utility> -#include <vector> #include <boost/container/static_vector.hpp> @@ -17,14 +13,12 @@ #include "video_core/rasterizer_accelerated.h" #include "video_core/rasterizer_interface.h" #include "video_core/renderer_vulkan/blit_image.h" -#include "video_core/renderer_vulkan/fixed_pipeline_state.h" #include "video_core/renderer_vulkan/vk_buffer_cache.h" #include "video_core/renderer_vulkan/vk_descriptor_pool.h" #include "video_core/renderer_vulkan/vk_fence_manager.h" #include "video_core/renderer_vulkan/vk_pipeline_cache.h" #include "video_core/renderer_vulkan/vk_query_cache.h" #include "video_core/renderer_vulkan/vk_render_pass_cache.h" -#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/renderer_vulkan/vk_update_descriptor.h" diff --git a/src/video_core/renderer_vulkan/vk_scheduler.cpp b/src/video_core/renderer_vulkan/vk_scheduler.cpp index 7d9d4f7ba..ad320991b 100644 --- a/src/video_core/renderer_vulkan/vk_scheduler.cpp +++ b/src/video_core/renderer_vulkan/vk_scheduler.cpp @@ -4,7 +4,6 @@ #include <memory> #include <mutex> -#include <optional> #include <thread> #include <utility> diff --git a/src/video_core/renderer_vulkan/vk_scheduler.h b/src/video_core/renderer_vulkan/vk_scheduler.h index e69aa136b..25c5e6ca1 100644 --- a/src/video_core/renderer_vulkan/vk_scheduler.h +++ b/src/video_core/renderer_vulkan/vk_scheduler.h @@ -4,7 +4,6 @@ #pragma once -#include <atomic> #include <condition_variable> #include <cstddef> #include <memory> diff --git a/src/video_core/renderer_vulkan/vk_shader_util.cpp b/src/video_core/renderer_vulkan/vk_shader_util.cpp index aaad4f292..e8e339f3c 100644 --- a/src/video_core/renderer_vulkan/vk_shader_util.cpp +++ b/src/video_core/renderer_vulkan/vk_shader_util.cpp @@ -3,9 +3,7 @@ // Refer to the license.txt file included. #include <cstring> -#include <memory> -#include "common/assert.h" #include "common/common_types.h" #include "video_core/renderer_vulkan/vk_shader_util.h" #include "video_core/vulkan_common/vulkan_device.h" diff --git a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp index 5d5329abf..64a58304b 100644 --- a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp +++ b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp @@ -118,7 +118,7 @@ StagingBufferPool::StagingBufferPool(const Device& device_, MemoryAllocator& mem .image = nullptr, .buffer = *stream_buffer, }; - const auto memory_properties = device.GetPhysical().GetMemoryProperties(); + const auto memory_properties = device.GetPhysical().GetMemoryProperties().memoryProperties; VkMemoryAllocateInfo stream_memory_info{ .sType = VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO, .pNext = make_dedicated ? &dedicated_info : nullptr, diff --git a/src/video_core/renderer_vulkan/vk_state_tracker.cpp b/src/video_core/renderer_vulkan/vk_state_tracker.cpp index c00913f55..1e597f98c 100644 --- a/src/video_core/renderer_vulkan/vk_state_tracker.cpp +++ b/src/video_core/renderer_vulkan/vk_state_tracker.cpp @@ -5,7 +5,6 @@ #include <algorithm> #include <array> #include <cstddef> -#include <iterator> #include "common/common_types.h" #include "core/core.h" diff --git a/src/video_core/renderer_vulkan/vk_swapchain.cpp b/src/video_core/renderer_vulkan/vk_swapchain.cpp index 8972a6921..ce744f4ca 100644 --- a/src/video_core/renderer_vulkan/vk_swapchain.cpp +++ b/src/video_core/renderer_vulkan/vk_swapchain.cpp @@ -7,11 +7,9 @@ #include <limits> #include <vector> -#include "common/assert.h" #include "common/logging/log.h" #include "common/settings.h" #include "core/core.h" -#include "core/frontend/framebuffer_layout.h" #include "video_core/renderer_vulkan/vk_scheduler.h" #include "video_core/renderer_vulkan/vk_swapchain.h" #include "video_core/vulkan_common/vulkan_device.h" diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.cpp b/src/video_core/renderer_vulkan/vk_texture_cache.cpp index 83a23b66a..2c2ccc7c6 100644 --- a/src/video_core/renderer_vulkan/vk_texture_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_texture_cache.cpp @@ -1189,6 +1189,14 @@ u64 TextureCacheRuntime::GetDeviceLocalMemory() const { return device.GetDeviceLocalMemory(); } +u64 TextureCacheRuntime::GetDeviceMemoryUsage() const { + return device.GetDeviceMemoryUsage(); +} + +bool TextureCacheRuntime::CanReportMemoryUsage() const { + return device.CanReportMemoryUsage(); +} + void TextureCacheRuntime::TickFrame() {} Image::Image(TextureCacheRuntime& runtime_, const ImageInfo& info_, GPUVAddr gpu_addr_, @@ -1203,6 +1211,7 @@ Image::Image(TextureCacheRuntime& runtime_, const ImageInfo& info_, GPUVAddr gpu } else { flags |= VideoCommon::ImageFlagBits::Converted; } + flags |= VideoCommon::ImageFlagBits::CostlyLoad; } if (runtime->device.HasDebuggingToolAttached()) { original_image.SetObjectNameEXT(VideoCommon::Name(*this).c_str()); @@ -1442,8 +1451,7 @@ bool Image::BlitScaleHelper(bool scale_up) { runtime->blit_image_helper.BlitColor(blit_framebuffer.get(), color_view, dst_region, src_region, operation, BLIT_OPERATION); - } else if (!runtime->device.IsBlitDepthStencilSupported() && - aspect_mask == (VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT)) { + } else if (aspect_mask == (VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT)) { if (!blit_framebuffer) { blit_framebuffer = std::make_unique<Framebuffer>(*runtime, nullptr, view_ptr, extent); } diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.h b/src/video_core/renderer_vulkan/vk_texture_cache.h index c81130dd2..cb15b4a1c 100644 --- a/src/video_core/renderer_vulkan/vk_texture_cache.h +++ b/src/video_core/renderer_vulkan/vk_texture_cache.h @@ -55,6 +55,10 @@ public: u64 GetDeviceLocalMemory() const; + u64 GetDeviceMemoryUsage() const; + + bool CanReportMemoryUsage() const; + void BlitImage(Framebuffer* dst_framebuffer, ImageView& dst, ImageView& src, const Region2D& dst_region, const Region2D& src_region, Tegra::Engines::Fermi2D::Filter filter, diff --git a/src/video_core/renderer_vulkan/vk_update_descriptor.cpp b/src/video_core/renderer_vulkan/vk_update_descriptor.cpp index 0df3a7fe9..89f1b508d 100644 --- a/src/video_core/renderer_vulkan/vk_update_descriptor.cpp +++ b/src/video_core/renderer_vulkan/vk_update_descriptor.cpp @@ -5,7 +5,6 @@ #include <variant> #include <boost/container/static_vector.hpp> -#include "common/assert.h" #include "common/logging/log.h" #include "video_core/renderer_vulkan/vk_scheduler.h" #include "video_core/renderer_vulkan/vk_update_descriptor.h" diff --git a/src/video_core/renderer_vulkan/vk_update_descriptor.h b/src/video_core/renderer_vulkan/vk_update_descriptor.h index d7de4c490..971a4eb34 100644 --- a/src/video_core/renderer_vulkan/vk_update_descriptor.h +++ b/src/video_core/renderer_vulkan/vk_update_descriptor.h @@ -6,7 +6,6 @@ #include <array> -#include "common/common_types.h" #include "video_core/vulkan_common/vulkan_wrapper.h" namespace Vulkan { diff --git a/src/video_core/shader_environment.cpp b/src/video_core/shader_environment.cpp index 3e673c437..3c22124c4 100644 --- a/src/video_core/shader_environment.cpp +++ b/src/video_core/shader_environment.cpp @@ -3,7 +3,6 @@ // Refer to the license.txt file included. #include <algorithm> -#include <bit> #include <filesystem> #include <fstream> #include <memory> diff --git a/src/video_core/shader_notify.cpp b/src/video_core/shader_notify.cpp index bcaf5f575..ce8925896 100644 --- a/src/video_core/shader_notify.cpp +++ b/src/video_core/shader_notify.cpp @@ -4,7 +4,6 @@ #include <atomic> #include <chrono> -#include <optional> #include "video_core/shader_notify.h" diff --git a/src/video_core/shader_notify.h b/src/video_core/shader_notify.h index 4d8d52071..538cda28a 100644 --- a/src/video_core/shader_notify.h +++ b/src/video_core/shader_notify.h @@ -6,7 +6,6 @@ #include <atomic> #include <chrono> -#include <optional> namespace VideoCore { class ShaderNotify { diff --git a/src/video_core/surface.cpp b/src/video_core/surface.cpp index a36015c8c..5f428d35d 100644 --- a/src/video_core/surface.cpp +++ b/src/video_core/surface.cpp @@ -190,13 +190,13 @@ PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format) } } -PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format) { +PixelFormat PixelFormatFromGPUPixelFormat(Service::android::PixelFormat format) { switch (format) { - case Tegra::FramebufferConfig::PixelFormat::A8B8G8R8_UNORM: + case Service::android::PixelFormat::Rgba8888: return PixelFormat::A8B8G8R8_UNORM; - case Tegra::FramebufferConfig::PixelFormat::RGB565_UNORM: + case Service::android::PixelFormat::Rgb565: return PixelFormat::R5G6B5_UNORM; - case Tegra::FramebufferConfig::PixelFormat::B8G8R8A8_UNORM: + case Service::android::PixelFormat::Bgra8888: return PixelFormat::B8G8R8A8_UNORM; default: UNIMPLEMENTED_MSG("Unimplemented format={}", format); diff --git a/src/video_core/surface.h b/src/video_core/surface.h index 33e8d24ab..5704cf16a 100644 --- a/src/video_core/surface.h +++ b/src/video_core/surface.h @@ -460,7 +460,7 @@ PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format); PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format); -PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat format); +PixelFormat PixelFormatFromGPUPixelFormat(Service::android::PixelFormat format); SurfaceType GetFormatType(PixelFormat pixel_format); diff --git a/src/video_core/texture_cache/descriptor_table.h b/src/video_core/texture_cache/descriptor_table.h index 3a03b786f..318bd5214 100644 --- a/src/video_core/texture_cache/descriptor_table.h +++ b/src/video_core/texture_cache/descriptor_table.h @@ -9,7 +9,6 @@ #include "common/common_types.h" #include "common/div_ceil.h" -#include "common/logging/log.h" #include "video_core/memory_manager.h" #include "video_core/rasterizer_interface.h" diff --git a/src/video_core/texture_cache/image_base.h b/src/video_core/texture_cache/image_base.h index 89c111c00..dd0106432 100644 --- a/src/video_core/texture_cache/image_base.h +++ b/src/video_core/texture_cache/image_base.h @@ -33,11 +33,12 @@ enum class ImageFlagBits : u32 { ///< garbage collection priority Alias = 1 << 11, ///< This image has aliases and has priority on garbage ///< collection + CostlyLoad = 1 << 12, ///< Protected from low-tier GC as it is costly to load back. // Rescaler - Rescaled = 1 << 12, - CheckingRescalable = 1 << 13, - IsRescalable = 1 << 14, + Rescaled = 1 << 13, + CheckingRescalable = 1 << 14, + IsRescalable = 1 << 15, }; DECLARE_ENUM_FLAG_OPERATORS(ImageFlagBits) diff --git a/src/video_core/texture_cache/render_targets.h b/src/video_core/texture_cache/render_targets.h index 0cb227d69..f8f13e84c 100644 --- a/src/video_core/texture_cache/render_targets.h +++ b/src/video_core/texture_cache/render_targets.h @@ -6,7 +6,6 @@ #include <algorithm> #include <span> -#include <utility> #include "common/bit_cast.h" #include "video_core/texture_cache/types.h" diff --git a/src/video_core/texture_cache/slot_vector.h b/src/video_core/texture_cache/slot_vector.h index 50df06409..6aabaef7b 100644 --- a/src/video_core/texture_cache/slot_vector.h +++ b/src/video_core/texture_cache/slot_vector.h @@ -5,9 +5,7 @@ #pragma once #include <algorithm> -#include <array> #include <bit> -#include <concepts> #include <numeric> #include <type_traits> #include <utility> diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h index 72eeb8bbd..efc1c4525 100644 --- a/src/video_core/texture_cache/texture_cache.h +++ b/src/video_core/texture_cache/texture_cache.h @@ -50,14 +50,20 @@ TextureCache<P>::TextureCache(Runtime& runtime_, VideoCore::RasterizerInterface& void(slot_samplers.insert(runtime, sampler_descriptor)); if constexpr (HAS_DEVICE_MEMORY_INFO) { - const auto device_memory = runtime.GetDeviceLocalMemory(); - const u64 possible_expected_memory = (device_memory * 4) / 10; - const u64 possible_critical_memory = (device_memory * 7) / 10; - expected_memory = std::max(possible_expected_memory, DEFAULT_EXPECTED_MEMORY - 256_MiB); - critical_memory = std::max(possible_critical_memory, DEFAULT_CRITICAL_MEMORY - 512_MiB); - minimum_memory = 0; + const s64 device_memory = static_cast<s64>(runtime.GetDeviceLocalMemory()); + const s64 min_spacing_expected = device_memory - 1_GiB - 512_MiB; + const s64 min_spacing_critical = device_memory - 1_GiB; + const s64 mem_threshold = std::min(device_memory, TARGET_THRESHOLD); + const s64 min_vacancy_expected = (6 * mem_threshold) / 10; + const s64 min_vacancy_critical = (3 * mem_threshold) / 10; + expected_memory = static_cast<u64>( + std::max(std::min(device_memory - min_vacancy_expected, min_spacing_expected), + DEFAULT_EXPECTED_MEMORY)); + critical_memory = static_cast<u64>( + std::max(std::min(device_memory - min_vacancy_critical, min_spacing_critical), + DEFAULT_CRITICAL_MEMORY)); + minimum_memory = static_cast<u64>((device_memory - mem_threshold) / 2); } else { - // On OpenGL we can be more conservatives as the driver takes care. expected_memory = DEFAULT_EXPECTED_MEMORY + 512_MiB; critical_memory = DEFAULT_CRITICAL_MEMORY + 1_GiB; minimum_memory = 0; @@ -66,18 +72,21 @@ TextureCache<P>::TextureCache(Runtime& runtime_, VideoCore::RasterizerInterface& template <class P> void TextureCache<P>::RunGarbageCollector() { - const bool high_priority_mode = total_used_memory >= expected_memory; - const bool aggressive_mode = total_used_memory >= critical_memory; - const u64 ticks_to_destroy = aggressive_mode ? 10ULL : high_priority_mode ? 25ULL : 100ULL; - size_t num_iterations = aggressive_mode ? 300 : (high_priority_mode ? 50 : 10); - const auto clean_up = [this, &num_iterations, high_priority_mode](ImageId image_id) { + bool high_priority_mode = total_used_memory >= expected_memory; + bool aggressive_mode = total_used_memory >= critical_memory; + const u64 ticks_to_destroy = aggressive_mode ? 10ULL : high_priority_mode ? 25ULL : 50ULL; + size_t num_iterations = aggressive_mode ? 40 : (high_priority_mode ? 20 : 10); + const auto clean_up = [this, &num_iterations, &high_priority_mode, + &aggressive_mode](ImageId image_id) { if (num_iterations == 0) { return true; } --num_iterations; auto& image = slot_images[image_id]; - const bool must_download = image.IsSafeDownload(); - if (!high_priority_mode && must_download) { + const bool must_download = + image.IsSafeDownload() && False(image.flags & ImageFlagBits::BadOverlap); + if (!high_priority_mode && + (must_download || True(image.flags & ImageFlagBits::CostlyLoad))) { return false; } if (must_download) { @@ -92,6 +101,18 @@ void TextureCache<P>::RunGarbageCollector() { } UnregisterImage(image_id); DeleteImage(image_id, image.scale_tick > frame_tick + 5); + if (total_used_memory < critical_memory) { + if (aggressive_mode) { + // Sink the aggresiveness. + num_iterations >>= 2; + aggressive_mode = false; + return false; + } + if (high_priority_mode && total_used_memory < expected_memory) { + num_iterations >>= 1; + high_priority_mode = false; + } + } return false; }; lru_cache.ForEachItemBelow(frame_tick - ticks_to_destroy, clean_up); @@ -99,6 +120,10 @@ void TextureCache<P>::RunGarbageCollector() { template <class P> 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(); + } if (total_used_memory > minimum_memory) { RunGarbageCollector(); } @@ -106,6 +131,7 @@ void TextureCache<P>::TickFrame() { sentenced_framebuffers.Tick(); sentenced_image_view.Tick(); runtime.TickFrame(); + critical_gc = 0; ++frame_tick; } @@ -1052,6 +1078,11 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA for (const ImageId overlap_id : overlap_ids) { Image& overlap = slot_images[overlap_id]; + if (True(overlap.flags & ImageFlagBits::GpuModified)) { + new_image.flags |= ImageFlagBits::GpuModified; + new_image.modification_tick = + std::max(overlap.modification_tick, new_image.modification_tick); + } if (overlap.info.num_samples != new_image.info.num_samples) { LOG_WARNING(HW_GPU, "Copying between images with different samples is not implemented"); } else { @@ -1414,6 +1445,10 @@ void TextureCache<P>::RegisterImage(ImageId image_id) { tentative_size = EstimatedDecompressedSize(tentative_size, image.info.format); } total_used_memory += Common::AlignUp(tentative_size, 1024); + if (total_used_memory > critical_memory && critical_gc < GC_EMERGENCY_COUNTS) { + RunGarbageCollector(); + critical_gc++; + } image.lru_index = lru_cache.Insert(image_id, frame_tick); ForEachGPUPage(image.gpu_addr, image.guest_size_bytes, @@ -1704,6 +1739,9 @@ void TextureCache<P>::SynchronizeAliases(ImageId image_id) { most_recent_tick = std::max(most_recent_tick, aliased_image.modification_tick); aliased_images.push_back(&aliased); any_rescaled |= True(aliased_image.flags & ImageFlagBits::Rescaled); + if (True(aliased_image.flags & ImageFlagBits::GpuModified)) { + image.flags |= ImageFlagBits::GpuModified; + } } } if (aliased_images.empty()) { diff --git a/src/video_core/texture_cache/texture_cache_base.h b/src/video_core/texture_cache/texture_cache_base.h index 7107887a6..b1324edf3 100644 --- a/src/video_core/texture_cache/texture_cache_base.h +++ b/src/video_core/texture_cache/texture_cache_base.h @@ -22,7 +22,6 @@ #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/slot_vector.h" #include "video_core/texture_cache/types.h" @@ -60,8 +59,10 @@ class TextureCache { /// 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; - static constexpr u64 DEFAULT_EXPECTED_MEMORY = 1_GiB; - static constexpr u64 DEFAULT_CRITICAL_MEMORY = 2_GiB; + static constexpr s64 TARGET_THRESHOLD = 4_GiB; + static constexpr s64 DEFAULT_EXPECTED_MEMORY = 1_GiB + 125_MiB; + static constexpr s64 DEFAULT_CRITICAL_MEMORY = 1_GiB + 625_MiB; + static constexpr size_t GC_EMERGENCY_COUNTS = 2; using Runtime = typename P::Runtime; using Image = typename P::Image; @@ -373,6 +374,7 @@ private: u64 minimum_memory; u64 expected_memory; u64 critical_memory; + size_t critical_gc; SlotVector<Image> slot_images; SlotVector<ImageMapView> slot_map_views; diff --git a/src/video_core/texture_cache/util.h b/src/video_core/texture_cache/util.h index 7af52de2e..f13669ea5 100644 --- a/src/video_core/texture_cache/util.h +++ b/src/video_core/texture_cache/util.h @@ -9,10 +9,8 @@ #include "common/common_types.h" -#include "video_core/engines/maxwell_3d.h" #include "video_core/surface.h" #include "video_core/texture_cache/image_base.h" -#include "video_core/texture_cache/image_view_base.h" #include "video_core/texture_cache/types.h" #include "video_core/textures/texture.h" diff --git a/src/video_core/textures/astc.cpp b/src/video_core/textures/astc.cpp index 25161df1f..28e4beafd 100644 --- a/src/video_core/textures/astc.cpp +++ b/src/video_core/textures/astc.cpp @@ -16,6 +16,7 @@ // <http://gamma.cs.unc.edu/FasTC/> #include <algorithm> +#include <bit> #include <cassert> #include <cstring> #include <span> diff --git a/src/video_core/textures/astc.h b/src/video_core/textures/astc.h index 14d2beec0..564ae1e36 100644 --- a/src/video_core/textures/astc.h +++ b/src/video_core/textures/astc.h @@ -4,9 +4,6 @@ #pragma once -#include <bit> -#include "common/common_types.h" - namespace Tegra::Texture::ASTC { void Decompress(std::span<const uint8_t> data, uint32_t width, uint32_t height, uint32_t depth, diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp index 24e943e4c..6dae23049 100644 --- a/src/video_core/textures/decoders.cpp +++ b/src/video_core/textures/decoders.cpp @@ -6,7 +6,6 @@ #include <cmath> #include <cstring> #include <span> -#include <utility> #include "common/alignment.h" #include "common/assert.h" @@ -14,7 +13,6 @@ #include "common/div_ceil.h" #include "video_core/gpu.h" #include "video_core/textures/decoders.h" -#include "video_core/textures/texture.h" namespace Tegra::Texture { namespace { diff --git a/src/video_core/textures/texture.cpp b/src/video_core/textures/texture.cpp index 06954963d..f56b4b9f9 100644 --- a/src/video_core/textures/texture.cpp +++ b/src/video_core/textures/texture.cpp @@ -2,7 +2,6 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include <algorithm> #include <array> #include "common/cityhash.h" diff --git a/src/video_core/vulkan_common/nsight_aftermath_tracker.h b/src/video_core/vulkan_common/nsight_aftermath_tracker.h index eae1891dd..55c115081 100644 --- a/src/video_core/vulkan_common/nsight_aftermath_tracker.h +++ b/src/video_core/vulkan_common/nsight_aftermath_tracker.h @@ -4,20 +4,22 @@ #pragma once -#include <filesystem> -#include <mutex> #include <span> -#include <string> -#include <vector> #include "common/common_types.h" -#include "common/dynamic_library.h" -#include "video_core/vulkan_common/vulkan_wrapper.h" #ifdef HAS_NSIGHT_AFTERMATH +#include <filesystem> +#include <mutex> + +// Vulkan headers must be included before Aftermath +#include "video_core/vulkan_common/vulkan_wrapper.h" + #include <GFSDK_Aftermath_Defines.h> #include <GFSDK_Aftermath_GpuCrashDump.h> #include <GFSDK_Aftermath_GpuCrashDumpDecoding.h> + +#include "common/dynamic_library.h" #endif namespace Vulkan { diff --git a/src/video_core/vulkan_common/vulkan_device.cpp b/src/video_core/vulkan_common/vulkan_device.cpp index effde73c9..4a5de9ddf 100644 --- a/src/video_core/vulkan_common/vulkan_device.cpp +++ b/src/video_core/vulkan_common/vulkan_device.cpp @@ -6,19 +6,20 @@ #include <bitset> #include <chrono> #include <optional> -#include <string_view> #include <thread> #include <unordered_set> #include <utility> #include <vector> #include "common/assert.h" +#include "common/literals.h" #include "common/settings.h" #include "video_core/vulkan_common/nsight_aftermath_tracker.h" #include "video_core/vulkan_common/vulkan_device.h" #include "video_core/vulkan_common/vulkan_wrapper.h" namespace Vulkan { +using namespace Common::Literals; namespace { namespace Alternatives { constexpr std::array STENCIL8_UINT{ @@ -597,6 +598,11 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR } logical = vk::Device::Create(physical, queue_cis, extensions, first_next, dld); + is_integrated = properties.deviceType == VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU; + is_virtual = properties.deviceType == VK_PHYSICAL_DEVICE_TYPE_VIRTUAL_GPU; + is_non_gpu = properties.deviceType == VK_PHYSICAL_DEVICE_TYPE_OTHER || + properties.deviceType == VK_PHYSICAL_DEVICE_TYPE_CPU; + CollectPhysicalMemoryInfo(); CollectTelemetryParameters(); CollectToolingInfo(); @@ -615,6 +621,11 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR khr_push_descriptor = false; break; } + const u32 nv_major_version = (properties.driverVersion >> 22) & 0x3ff; + if (nv_major_version >= 510) { + LOG_WARNING(Render_Vulkan, "NVIDIA Drivers >= 510 do not support MSAA image blits"); + cant_blit_msaa = true; + } } const bool is_radv = driver_id == VK_DRIVER_ID_MESA_RADV; if (ext_extended_dynamic_state && is_radv) { @@ -986,6 +997,7 @@ std::vector<const char*> Device::LoadExtensions(bool requires_surface) { test(has_khr_swapchain_mutable_format, VK_KHR_SWAPCHAIN_MUTABLE_FORMAT_EXTENSION_NAME, false); test(has_ext_line_rasterization, VK_EXT_LINE_RASTERIZATION_EXTENSION_NAME, false); + test(ext_memory_budget, VK_EXT_MEMORY_BUDGET_EXTENSION_NAME, true); if (Settings::values.enable_nsight_aftermath) { test(nv_device_diagnostics_config, VK_NV_DEVICE_DIAGNOSTICS_CONFIG_EXTENSION_NAME, true); @@ -998,7 +1010,7 @@ std::vector<const char*> Device::LoadExtensions(bool requires_surface) { VkPhysicalDeviceFeatures2KHR features{}; features.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_FEATURES_2_KHR; - VkPhysicalDeviceProperties2KHR physical_properties; + VkPhysicalDeviceProperties2KHR physical_properties{}; physical_properties.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROPERTIES_2_KHR; if (has_khr_shader_float16_int8) { @@ -1268,15 +1280,50 @@ void Device::CollectTelemetryParameters() { vendor_name = driver.driverName; } +u64 Device::GetDeviceMemoryUsage() const { + VkPhysicalDeviceMemoryBudgetPropertiesEXT budget; + budget.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MEMORY_BUDGET_PROPERTIES_EXT; + budget.pNext = nullptr; + physical.GetMemoryProperties(&budget); + u64 result{}; + for (const size_t heap : valid_heap_memory) { + result += budget.heapUsage[heap]; + } + return result; +} + void Device::CollectPhysicalMemoryInfo() { - const auto mem_properties = physical.GetMemoryProperties(); + VkPhysicalDeviceMemoryBudgetPropertiesEXT budget{}; + budget.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MEMORY_BUDGET_PROPERTIES_EXT; + const auto mem_info = physical.GetMemoryProperties(ext_memory_budget ? &budget : nullptr); + const auto& mem_properties = mem_info.memoryProperties; const size_t num_properties = mem_properties.memoryHeapCount; device_access_memory = 0; + u64 device_initial_usage = 0; + u64 local_memory = 0; for (size_t element = 0; element < num_properties; ++element) { - if ((mem_properties.memoryHeaps[element].flags & VK_MEMORY_HEAP_DEVICE_LOCAL_BIT) != 0) { - device_access_memory += mem_properties.memoryHeaps[element].size; + const bool is_heap_local = + (mem_properties.memoryHeaps[element].flags & VK_MEMORY_HEAP_DEVICE_LOCAL_BIT) != 0; + if (!is_integrated && !is_heap_local) { + continue; } + valid_heap_memory.push_back(element); + if (is_heap_local) { + local_memory += mem_properties.memoryHeaps[element].size; + } + if (ext_memory_budget) { + device_initial_usage += budget.heapUsage[element]; + device_access_memory += budget.heapBudget[element]; + continue; + } + device_access_memory += mem_properties.memoryHeaps[element].size; + } + if (!is_integrated) { + return; } + const s64 available_memory = static_cast<s64>(device_access_memory - device_initial_usage); + device_access_memory = static_cast<u64>(std::max<s64>( + std::min<s64>(available_memory - 8_GiB, 4_GiB), static_cast<s64>(local_memory))); } void Device::CollectToolingInfo() { diff --git a/src/video_core/vulkan_common/vulkan_device.h b/src/video_core/vulkan_common/vulkan_device.h index 34b1add16..2d709d069 100644 --- a/src/video_core/vulkan_common/vulkan_device.h +++ b/src/video_core/vulkan_common/vulkan_device.h @@ -6,7 +6,6 @@ #include <span> #include <string> -#include <string_view> #include <unordered_map> #include <vector> @@ -342,6 +341,12 @@ public: return device_access_memory; } + bool CanReportMemoryUsage() const { + return ext_memory_budget; + } + + u64 GetDeviceMemoryUsage() const; + u32 GetSetsPerPool() const { return sets_per_pool; } @@ -422,6 +427,9 @@ private: bool is_topology_list_restart_supported{}; ///< Support for primitive restart with list ///< topologies. bool is_patch_list_restart_supported{}; ///< Support for primitive restart with list patch. + bool is_integrated{}; ///< Is GPU an iGPU. + bool is_virtual{}; ///< Is GPU a virtual GPU. + bool is_non_gpu{}; ///< Is SoftwareRasterizer, FPGA, non-GPU device. bool nv_viewport_swizzle{}; ///< Support for VK_NV_viewport_swizzle. bool nv_viewport_array2{}; ///< Support for VK_NV_viewport_array2. bool nv_geometry_shader_passthrough{}; ///< Support for VK_NV_geometry_shader_passthrough. @@ -446,6 +454,7 @@ private: bool ext_shader_atomic_int64{}; ///< Support for VK_KHR_shader_atomic_int64. bool ext_conservative_rasterization{}; ///< Support for VK_EXT_conservative_rasterization. bool ext_provoking_vertex{}; ///< Support for VK_EXT_provoking_vertex. + bool ext_memory_budget{}; ///< Support for VK_EXT_memory_budget. bool nv_device_diagnostics_config{}; ///< Support for VK_NV_device_diagnostics_config. bool has_broken_cube_compatibility{}; ///< Has broken cube compatiblity bit bool has_renderdoc{}; ///< Has RenderDoc attached @@ -457,6 +466,7 @@ private: // Telemetry parameters std::string vendor_name; ///< Device's driver name. std::vector<std::string> supported_extensions; ///< Reported Vulkan extensions. + std::vector<size_t> valid_heap_memory; ///< Heaps used. /// Format properties dictionary. std::unordered_map<VkFormat, VkFormatProperties> format_properties; diff --git a/src/video_core/vulkan_common/vulkan_instance.cpp b/src/video_core/vulkan_common/vulkan_instance.cpp index bfd6e6add..662694f16 100644 --- a/src/video_core/vulkan_common/vulkan_instance.cpp +++ b/src/video_core/vulkan_common/vulkan_instance.cpp @@ -2,11 +2,9 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include <algorithm> #include <future> #include <optional> #include <span> -#include <utility> #include <vector> #include "common/common_types.h" diff --git a/src/video_core/vulkan_common/vulkan_library.cpp b/src/video_core/vulkan_common/vulkan_library.cpp index 22833fa56..d69de05ef 100644 --- a/src/video_core/vulkan_common/vulkan_library.cpp +++ b/src/video_core/vulkan_common/vulkan_library.cpp @@ -2,7 +2,6 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include <cstdlib> #include <string> #include "common/dynamic_library.h" diff --git a/src/video_core/vulkan_common/vulkan_memory_allocator.cpp b/src/video_core/vulkan_common/vulkan_memory_allocator.cpp index 300a61205..e6e97b332 100644 --- a/src/video_core/vulkan_common/vulkan_memory_allocator.cpp +++ b/src/video_core/vulkan_common/vulkan_memory_allocator.cpp @@ -227,7 +227,7 @@ void MemoryCommit::Release() { } MemoryAllocator::MemoryAllocator(const Device& device_, bool export_allocations_) - : device{device_}, properties{device_.GetPhysical().GetMemoryProperties()}, + : device{device_}, properties{device_.GetPhysical().GetMemoryProperties().memoryProperties}, export_allocations{export_allocations_}, buffer_image_granularity{ device_.GetPhysical().GetProperties().limits.bufferImageGranularity} {} diff --git a/src/video_core/vulkan_common/vulkan_memory_allocator.h b/src/video_core/vulkan_common/vulkan_memory_allocator.h index 86e8ed119..338daf5ba 100644 --- a/src/video_core/vulkan_common/vulkan_memory_allocator.h +++ b/src/video_core/vulkan_common/vulkan_memory_allocator.h @@ -6,7 +6,6 @@ #include <memory> #include <span> -#include <utility> #include <vector> #include "common/common_types.h" #include "video_core/vulkan_common/vulkan_wrapper.h" diff --git a/src/video_core/vulkan_common/vulkan_wrapper.cpp b/src/video_core/vulkan_common/vulkan_wrapper.cpp index a9faa4807..742cc39da 100644 --- a/src/video_core/vulkan_common/vulkan_wrapper.cpp +++ b/src/video_core/vulkan_common/vulkan_wrapper.cpp @@ -3,10 +3,8 @@ // Refer to the license.txt file included. #include <algorithm> -#include <exception> #include <memory> #include <optional> -#include <string_view> #include <utility> #include <vector> @@ -239,8 +237,8 @@ bool Load(VkInstance instance, InstanceDispatch& dld) noexcept { return X(vkCreateDevice) && X(vkDestroyDevice) && X(vkDestroyDevice) && X(vkEnumerateDeviceExtensionProperties) && X(vkEnumeratePhysicalDevices) && X(vkGetDeviceProcAddr) && X(vkGetPhysicalDeviceFormatProperties) && - X(vkGetPhysicalDeviceMemoryProperties) && X(vkGetPhysicalDeviceProperties) && - X(vkGetPhysicalDeviceQueueFamilyProperties); + X(vkGetPhysicalDeviceMemoryProperties) && X(vkGetPhysicalDeviceMemoryProperties2) && + X(vkGetPhysicalDeviceProperties) && X(vkGetPhysicalDeviceQueueFamilyProperties); #undef X } @@ -928,9 +926,12 @@ std::vector<VkPresentModeKHR> PhysicalDevice::GetSurfacePresentModesKHR( return modes; } -VkPhysicalDeviceMemoryProperties PhysicalDevice::GetMemoryProperties() const noexcept { - VkPhysicalDeviceMemoryProperties properties; - dld->vkGetPhysicalDeviceMemoryProperties(physical_device, &properties); +VkPhysicalDeviceMemoryProperties2 PhysicalDevice::GetMemoryProperties( + void* next_structures) const noexcept { + VkPhysicalDeviceMemoryProperties2 properties{}; + properties.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MEMORY_PROPERTIES_2; + properties.pNext = next_structures; + dld->vkGetPhysicalDeviceMemoryProperties2(physical_device, &properties); return properties; } diff --git a/src/video_core/vulkan_common/vulkan_wrapper.h b/src/video_core/vulkan_common/vulkan_wrapper.h index b7ae01c6c..0a5f9931c 100644 --- a/src/video_core/vulkan_common/vulkan_wrapper.h +++ b/src/video_core/vulkan_common/vulkan_wrapper.h @@ -5,7 +5,6 @@ #pragma once #include <exception> -#include <iterator> #include <limits> #include <memory> #include <optional> @@ -173,6 +172,7 @@ struct InstanceDispatch { PFN_vkGetPhysicalDeviceFeatures2KHR vkGetPhysicalDeviceFeatures2KHR{}; PFN_vkGetPhysicalDeviceFormatProperties vkGetPhysicalDeviceFormatProperties{}; PFN_vkGetPhysicalDeviceMemoryProperties vkGetPhysicalDeviceMemoryProperties{}; + PFN_vkGetPhysicalDeviceMemoryProperties2 vkGetPhysicalDeviceMemoryProperties2{}; PFN_vkGetPhysicalDeviceProperties vkGetPhysicalDeviceProperties{}; PFN_vkGetPhysicalDeviceProperties2KHR vkGetPhysicalDeviceProperties2KHR{}; PFN_vkGetPhysicalDeviceQueueFamilyProperties vkGetPhysicalDeviceQueueFamilyProperties{}; @@ -951,7 +951,8 @@ public: std::vector<VkPresentModeKHR> GetSurfacePresentModesKHR(VkSurfaceKHR) const; - VkPhysicalDeviceMemoryProperties GetMemoryProperties() const noexcept; + VkPhysicalDeviceMemoryProperties2 GetMemoryProperties( + void* next_structures = nullptr) const noexcept; private: VkPhysicalDevice physical_device = nullptr; |