From 9568b310befe19154511afe14709188f641e4951 Mon Sep 17 00:00:00 2001 From: Liam Date: Sat, 13 Jan 2024 22:52:04 -0500 Subject: renderer_vulkan: isolate FXAA from blit screen --- .../renderer_vulkan/present/anti_alias_pass.h | 27 ++ src/video_core/renderer_vulkan/present/fsr.cpp | 420 ++++++++++++++++++ src/video_core/renderer_vulkan/present/fsr.h | 52 +++ src/video_core/renderer_vulkan/present/fxaa.cpp | 144 ++++++ src/video_core/renderer_vulkan/present/fxaa.h | 63 +++ src/video_core/renderer_vulkan/present/smaa.cpp | 270 ++++++++++++ src/video_core/renderer_vulkan/present/smaa.h | 87 ++++ src/video_core/renderer_vulkan/present/util.cpp | 488 +++++++++++++++++++++ src/video_core/renderer_vulkan/present/util.h | 46 ++ 9 files changed, 1597 insertions(+) create mode 100644 src/video_core/renderer_vulkan/present/anti_alias_pass.h create mode 100644 src/video_core/renderer_vulkan/present/fsr.cpp create mode 100644 src/video_core/renderer_vulkan/present/fsr.h create mode 100644 src/video_core/renderer_vulkan/present/fxaa.cpp create mode 100644 src/video_core/renderer_vulkan/present/fxaa.h create mode 100644 src/video_core/renderer_vulkan/present/smaa.cpp create mode 100644 src/video_core/renderer_vulkan/present/smaa.h create mode 100644 src/video_core/renderer_vulkan/present/util.cpp create mode 100644 src/video_core/renderer_vulkan/present/util.h (limited to 'src/video_core/renderer_vulkan/present') diff --git a/src/video_core/renderer_vulkan/present/anti_alias_pass.h b/src/video_core/renderer_vulkan/present/anti_alias_pass.h new file mode 100644 index 000000000..c1ec0b9a0 --- /dev/null +++ b/src/video_core/renderer_vulkan/present/anti_alias_pass.h @@ -0,0 +1,27 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include "video_core/vulkan_common/vulkan_wrapper.h" + +namespace Vulkan { + +class Scheduler; + +class AntiAliasPass { +public: + virtual ~AntiAliasPass() = default; + virtual VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, + VkImageView source_image_view) = 0; +}; + +class NoAA final : public AntiAliasPass { +public: + virtual VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, + VkImageView source_image_view) { + return source_image_view; + } +}; + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/fsr.cpp b/src/video_core/renderer_vulkan/present/fsr.cpp new file mode 100644 index 000000000..30a16a785 --- /dev/null +++ b/src/video_core/renderer_vulkan/present/fsr.cpp @@ -0,0 +1,420 @@ +// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#include "common/common_types.h" +#include "common/div_ceil.h" +#include "common/settings.h" + +#include "video_core/fsr.h" +#include "video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp16_comp_spv.h" +#include "video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp32_comp_spv.h" +#include "video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp16_comp_spv.h" +#include "video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp32_comp_spv.h" +#include "video_core/renderer_vulkan/present/fsr.h" +#include "video_core/renderer_vulkan/vk_scheduler.h" +#include "video_core/renderer_vulkan/vk_shader_util.h" +#include "video_core/vulkan_common/vulkan_device.h" + +namespace Vulkan { +using namespace FSR; + +FSR::FSR(const Device& device_, MemoryAllocator& memory_allocator_, size_t image_count_, + VkExtent2D output_size_) + : device{device_}, memory_allocator{memory_allocator_}, image_count{image_count_}, + output_size{output_size_} { + + CreateImages(); + CreateSampler(); + CreateShaders(); + CreateDescriptorPool(); + CreateDescriptorSetLayout(); + CreateDescriptorSets(); + CreatePipelineLayout(); + CreatePipeline(); +} + +VkImageView FSR::Draw(Scheduler& scheduler, size_t image_index, VkImageView image_view, + VkExtent2D input_image_extent, const Common::Rectangle& crop_rect) { + + UpdateDescriptorSet(image_index, image_view); + + scheduler.Record([this, image_index, input_image_extent, crop_rect](vk::CommandBuffer cmdbuf) { + const VkImageMemoryBarrier base_barrier{ + .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER, + .pNext = nullptr, + .srcAccessMask = 0, + .dstAccessMask = 0, + .oldLayout = VK_IMAGE_LAYOUT_GENERAL, + .newLayout = VK_IMAGE_LAYOUT_GENERAL, + .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, + .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, + .image = {}, + .subresourceRange = + { + .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .baseMipLevel = 0, + .levelCount = 1, + .baseArrayLayer = 0, + .layerCount = 1, + }, + }; + + cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, *easu_pipeline); + + const f32 input_image_width = static_cast(input_image_extent.width); + const f32 input_image_height = static_cast(input_image_extent.height); + const f32 output_image_width = static_cast(output_size.width); + const f32 output_image_height = static_cast(output_size.height); + const f32 viewport_width = (crop_rect.right - crop_rect.left) * input_image_width; + const f32 viewport_x = crop_rect.left * input_image_width; + const f32 viewport_height = (crop_rect.bottom - crop_rect.top) * input_image_height; + const f32 viewport_y = crop_rect.top * input_image_height; + + std::array push_constants; + FsrEasuConOffset(push_constants.data() + 0, push_constants.data() + 4, + push_constants.data() + 8, push_constants.data() + 12, + + viewport_width, viewport_height, input_image_width, input_image_height, + output_image_width, output_image_height, viewport_x, viewport_y); + cmdbuf.PushConstants(*pipeline_layout, VK_SHADER_STAGE_COMPUTE_BIT, push_constants); + + { + VkImageMemoryBarrier fsr_write_barrier = base_barrier; + fsr_write_barrier.image = *images[image_index]; + fsr_write_barrier.oldLayout = VK_IMAGE_LAYOUT_UNDEFINED; + + cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, + VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0, fsr_write_barrier); + } + + cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, *pipeline_layout, 0, + descriptor_sets[image_index * 2], {}); + cmdbuf.Dispatch(Common::DivCeil(output_size.width, 16u), + Common::DivCeil(output_size.height, 16u), 1); + + cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, *rcas_pipeline); + + const float sharpening = + static_cast(Settings::values.fsr_sharpening_slider.GetValue()) / 100.0f; + + FsrRcasCon(push_constants.data(), sharpening); + cmdbuf.PushConstants(*pipeline_layout, VK_SHADER_STAGE_COMPUTE_BIT, push_constants); + + { + std::array barriers; + auto& fsr_read_barrier = barriers[0]; + auto& blit_write_barrier = barriers[1]; + + fsr_read_barrier = base_barrier; + fsr_read_barrier.image = *images[image_index]; + fsr_read_barrier.srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT; + fsr_read_barrier.dstAccessMask = VK_ACCESS_SHADER_READ_BIT; + + blit_write_barrier = base_barrier; + blit_write_barrier.image = *images[image_count + image_index]; + blit_write_barrier.oldLayout = VK_IMAGE_LAYOUT_UNDEFINED; + blit_write_barrier.newLayout = VK_IMAGE_LAYOUT_GENERAL; + + cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, + VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0, {}, {}, barriers); + } + + cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, *pipeline_layout, 0, + descriptor_sets[image_index * 2 + 1], {}); + cmdbuf.Dispatch(Common::DivCeil(output_size.width, 16u), + Common::DivCeil(output_size.height, 16u), 1); + + { + std::array barriers; + auto& blit_read_barrier = barriers[0]; + + blit_read_barrier = base_barrier; + blit_read_barrier.image = *images[image_count + image_index]; + blit_read_barrier.srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT; + blit_read_barrier.dstAccessMask = VK_ACCESS_SHADER_READ_BIT; + + cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, + VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT, 0, {}, {}, barriers); + } + }); + + return *image_views[image_count + image_index]; +} + +void FSR::CreateDescriptorPool() { + const std::array pool_sizes{{ + { + .type = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + .descriptorCount = static_cast(image_count * 2), + }, + { + .type = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE, + .descriptorCount = static_cast(image_count * 2), + }, + }}; + + const VkDescriptorPoolCreateInfo ci{ + .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .maxSets = static_cast(image_count * 2), + .poolSizeCount = static_cast(pool_sizes.size()), + .pPoolSizes = pool_sizes.data(), + }; + descriptor_pool = device.GetLogical().CreateDescriptorPool(ci); +} + +void FSR::CreateDescriptorSetLayout() { + const std::array layout_bindings{{ + { + .binding = 0, + .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + .descriptorCount = 1, + .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT, + .pImmutableSamplers = sampler.address(), + }, + { + .binding = 1, + .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE, + .descriptorCount = 1, + .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT, + .pImmutableSamplers = sampler.address(), + }, + }}; + + const VkDescriptorSetLayoutCreateInfo ci{ + .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .bindingCount = static_cast(layout_bindings.size()), + .pBindings = layout_bindings.data(), + }; + + descriptor_set_layout = device.GetLogical().CreateDescriptorSetLayout(ci); +} + +void FSR::CreateDescriptorSets() { + const u32 sets = static_cast(image_count * 2); + const std::vector layouts(sets, *descriptor_set_layout); + + const VkDescriptorSetAllocateInfo ai{ + .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_ALLOCATE_INFO, + .pNext = nullptr, + .descriptorPool = *descriptor_pool, + .descriptorSetCount = sets, + .pSetLayouts = layouts.data(), + }; + + descriptor_sets = descriptor_pool.Allocate(ai); +} + +void FSR::CreateImages() { + images.resize(image_count * 2); + image_views.resize(image_count * 2); + + for (size_t i = 0; i < image_count * 2; ++i) { + images[i] = memory_allocator.CreateImage(VkImageCreateInfo{ + .sType = VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .imageType = VK_IMAGE_TYPE_2D, + .format = VK_FORMAT_R16G16B16A16_SFLOAT, + .extent = + { + .width = output_size.width, + .height = output_size.height, + .depth = 1, + }, + .mipLevels = 1, + .arrayLayers = 1, + .samples = VK_SAMPLE_COUNT_1_BIT, + .tiling = VK_IMAGE_TILING_OPTIMAL, + .usage = VK_IMAGE_USAGE_TRANSFER_SRC_BIT | VK_IMAGE_USAGE_STORAGE_BIT | + VK_IMAGE_USAGE_SAMPLED_BIT, + .sharingMode = VK_SHARING_MODE_EXCLUSIVE, + .queueFamilyIndexCount = 0, + .pQueueFamilyIndices = nullptr, + .initialLayout = VK_IMAGE_LAYOUT_UNDEFINED, + }); + image_views[i] = device.GetLogical().CreateImageView(VkImageViewCreateInfo{ + .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .image = *images[i], + .viewType = VK_IMAGE_VIEW_TYPE_2D, + .format = VK_FORMAT_R16G16B16A16_SFLOAT, + .components = + { + .r = VK_COMPONENT_SWIZZLE_IDENTITY, + .g = VK_COMPONENT_SWIZZLE_IDENTITY, + .b = VK_COMPONENT_SWIZZLE_IDENTITY, + .a = VK_COMPONENT_SWIZZLE_IDENTITY, + }, + .subresourceRange = + { + .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .baseMipLevel = 0, + .levelCount = 1, + .baseArrayLayer = 0, + .layerCount = 1, + }, + }); + } +} + +void FSR::CreatePipelineLayout() { + VkPushConstantRange push_const{ + .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT, + .offset = 0, + .size = sizeof(std::array), + }; + VkPipelineLayoutCreateInfo ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .setLayoutCount = 1, + .pSetLayouts = descriptor_set_layout.address(), + .pushConstantRangeCount = 1, + .pPushConstantRanges = &push_const, + }; + + pipeline_layout = device.GetLogical().CreatePipelineLayout(ci); +} + +void FSR::UpdateDescriptorSet(std::size_t image_index, VkImageView image_view) const { + const auto fsr_image_view = *image_views[image_index]; + const auto blit_image_view = *image_views[image_count + image_index]; + + const VkDescriptorImageInfo image_info{ + .sampler = VK_NULL_HANDLE, + .imageView = image_view, + .imageLayout = VK_IMAGE_LAYOUT_GENERAL, + }; + const VkDescriptorImageInfo fsr_image_info{ + .sampler = VK_NULL_HANDLE, + .imageView = fsr_image_view, + .imageLayout = VK_IMAGE_LAYOUT_GENERAL, + }; + const VkDescriptorImageInfo blit_image_info{ + .sampler = VK_NULL_HANDLE, + .imageView = blit_image_view, + .imageLayout = VK_IMAGE_LAYOUT_GENERAL, + }; + + VkWriteDescriptorSet sampler_write{ + .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET, + .pNext = nullptr, + .dstSet = descriptor_sets[image_index * 2], + .dstBinding = 0, + .dstArrayElement = 0, + .descriptorCount = 1, + .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + .pImageInfo = &image_info, + .pBufferInfo = nullptr, + .pTexelBufferView = nullptr, + }; + + VkWriteDescriptorSet output_write{ + .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET, + .pNext = nullptr, + .dstSet = descriptor_sets[image_index * 2], + .dstBinding = 1, + .dstArrayElement = 0, + .descriptorCount = 1, + .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE, + .pImageInfo = &fsr_image_info, + .pBufferInfo = nullptr, + .pTexelBufferView = nullptr, + }; + + device.GetLogical().UpdateDescriptorSets(std::array{sampler_write, output_write}, {}); + + sampler_write.dstSet = descriptor_sets[image_index * 2 + 1]; + sampler_write.pImageInfo = &fsr_image_info; + output_write.dstSet = descriptor_sets[image_index * 2 + 1]; + output_write.pImageInfo = &blit_image_info; + + device.GetLogical().UpdateDescriptorSets(std::array{sampler_write, output_write}, {}); +} + +void FSR::CreateSampler() { + const VkSamplerCreateInfo ci{ + .sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .magFilter = VK_FILTER_LINEAR, + .minFilter = VK_FILTER_LINEAR, + .mipmapMode = VK_SAMPLER_MIPMAP_MODE_LINEAR, + .addressModeU = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE, + .addressModeV = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE, + .addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE, + .mipLodBias = 0.0f, + .anisotropyEnable = VK_FALSE, + .maxAnisotropy = 0.0f, + .compareEnable = VK_FALSE, + .compareOp = VK_COMPARE_OP_NEVER, + .minLod = 0.0f, + .maxLod = 0.0f, + .borderColor = VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK, + .unnormalizedCoordinates = VK_FALSE, + }; + + sampler = device.GetLogical().CreateSampler(ci); +} + +void FSR::CreateShaders() { + if (device.IsFloat16Supported()) { + easu_shader = BuildShader(device, VULKAN_FIDELITYFX_FSR_EASU_FP16_COMP_SPV); + rcas_shader = BuildShader(device, VULKAN_FIDELITYFX_FSR_RCAS_FP16_COMP_SPV); + } else { + easu_shader = BuildShader(device, VULKAN_FIDELITYFX_FSR_EASU_FP32_COMP_SPV); + rcas_shader = BuildShader(device, VULKAN_FIDELITYFX_FSR_RCAS_FP32_COMP_SPV); + } +} + +void FSR::CreatePipeline() { + VkPipelineShaderStageCreateInfo shader_stage_easu{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .stage = VK_SHADER_STAGE_COMPUTE_BIT, + .module = *easu_shader, + .pName = "main", + .pSpecializationInfo = nullptr, + }; + + VkPipelineShaderStageCreateInfo shader_stage_rcas{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .stage = VK_SHADER_STAGE_COMPUTE_BIT, + .module = *rcas_shader, + .pName = "main", + .pSpecializationInfo = nullptr, + }; + + VkComputePipelineCreateInfo pipeline_ci_easu{ + .sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .stage = shader_stage_easu, + .layout = *pipeline_layout, + .basePipelineHandle = VK_NULL_HANDLE, + .basePipelineIndex = 0, + }; + + VkComputePipelineCreateInfo pipeline_ci_rcas{ + .sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .stage = shader_stage_rcas, + .layout = *pipeline_layout, + .basePipelineHandle = VK_NULL_HANDLE, + .basePipelineIndex = 0, + }; + + easu_pipeline = device.GetLogical().CreateComputePipeline(pipeline_ci_easu); + rcas_pipeline = device.GetLogical().CreateComputePipeline(pipeline_ci_rcas); +} + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/fsr.h b/src/video_core/renderer_vulkan/present/fsr.h new file mode 100644 index 000000000..3505c1416 --- /dev/null +++ b/src/video_core/renderer_vulkan/present/fsr.h @@ -0,0 +1,52 @@ +// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include "common/math_util.h" +#include "video_core/vulkan_common/vulkan_memory_allocator.h" +#include "video_core/vulkan_common/vulkan_wrapper.h" + +namespace Vulkan { + +class Device; +class Scheduler; + +class FSR { +public: + explicit FSR(const Device& device, MemoryAllocator& memory_allocator, size_t image_count, + VkExtent2D output_size); + VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImageView image_view, + VkExtent2D input_image_extent, const Common::Rectangle& crop_rect); + +private: + void CreateDescriptorPool(); + void CreateDescriptorSetLayout(); + void CreateDescriptorSets(); + void CreateImages(); + void CreateSampler(); + void CreateShaders(); + void CreatePipeline(); + void CreatePipelineLayout(); + + void UpdateDescriptorSet(std::size_t image_index, VkImageView image_view) const; + + const Device& device; + MemoryAllocator& memory_allocator; + size_t image_count; + VkExtent2D output_size; + + vk::DescriptorPool descriptor_pool; + vk::DescriptorSetLayout descriptor_set_layout; + vk::DescriptorSets descriptor_sets; + vk::PipelineLayout pipeline_layout; + vk::ShaderModule easu_shader; + vk::ShaderModule rcas_shader; + vk::Pipeline easu_pipeline; + vk::Pipeline rcas_pipeline; + vk::Sampler sampler; + std::vector images; + std::vector image_views; +}; + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/fxaa.cpp b/src/video_core/renderer_vulkan/present/fxaa.cpp new file mode 100644 index 000000000..6f87ddebb --- /dev/null +++ b/src/video_core/renderer_vulkan/present/fxaa.cpp @@ -0,0 +1,144 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#include "common/common_types.h" + +#include "video_core/host_shaders/fxaa_frag_spv.h" +#include "video_core/host_shaders/fxaa_vert_spv.h" +#include "video_core/renderer_vulkan/present/fxaa.h" +#include "video_core/renderer_vulkan/present/util.h" +#include "video_core/renderer_vulkan/vk_scheduler.h" +#include "video_core/renderer_vulkan/vk_shader_util.h" +#include "video_core/vulkan_common/vulkan_device.h" + +namespace Vulkan { + +FXAA::FXAA(const Device& device, MemoryAllocator& allocator, size_t image_count, VkExtent2D extent) + : m_device(device), m_allocator(allocator), m_extent(extent), + m_image_count(static_cast(image_count)) { + CreateImages(); + CreateRenderPasses(); + CreateSampler(); + CreateShaders(); + CreateDescriptorPool(); + CreateDescriptorSetLayouts(); + CreateDescriptorSets(); + CreatePipelineLayouts(); + CreatePipelines(); +} + +FXAA::~FXAA() = default; + +void FXAA::CreateImages() { + for (u32 i = 0; i < m_image_count; i++) { + Image& image = m_dynamic_images.emplace_back(); + + image.image = CreateWrappedImage(m_allocator, m_extent, VK_FORMAT_R16G16B16A16_SFLOAT); + image.image_view = + CreateWrappedImageView(m_device, image.image, VK_FORMAT_R16G16B16A16_SFLOAT); + } +} + +void FXAA::CreateRenderPasses() { + m_renderpass = CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16B16A16_SFLOAT); + + for (auto& image : m_dynamic_images) { + image.framebuffer = + CreateWrappedFramebuffer(m_device, m_renderpass, image.image_view, m_extent); + } +} + +void FXAA::CreateSampler() { + m_sampler = CreateWrappedSampler(m_device); +} + +void FXAA::CreateShaders() { + m_vertex_shader = CreateWrappedShaderModule(m_device, FXAA_VERT_SPV); + m_fragment_shader = CreateWrappedShaderModule(m_device, FXAA_FRAG_SPV); +} + +void FXAA::CreateDescriptorPool() { + // 2 descriptors, 1 descriptor set per image + m_descriptor_pool = CreateWrappedDescriptorPool(m_device, 2 * m_image_count, m_image_count); +} + +void FXAA::CreateDescriptorSetLayouts() { + m_descriptor_set_layout = CreateWrappedDescriptorSetLayout(m_device, 2); +} + +void FXAA::CreateDescriptorSets() { + VkDescriptorSetLayout layout = *m_descriptor_set_layout; + + for (auto& images : m_dynamic_images) { + images.descriptor_sets = CreateWrappedDescriptorSets(m_descriptor_pool, {layout}); + } +} + +void FXAA::CreatePipelineLayouts() { + m_pipeline_layout = CreateWrappedPipelineLayout(m_device, m_descriptor_set_layout); +} + +void FXAA::CreatePipelines() { + m_pipeline = CreateWrappedPipeline(m_device, m_renderpass, m_pipeline_layout, + std::tie(m_vertex_shader, m_fragment_shader)); +} + +void FXAA::UpdateDescriptorSets(VkImageView image_view, size_t image_index) { + Image& image = m_dynamic_images[image_index]; + std::vector image_infos; + std::vector updates; + image_infos.reserve(2); + + updates.push_back( + CreateWriteDescriptorSet(image_infos, *m_sampler, image_view, image.descriptor_sets[0], 0)); + updates.push_back( + CreateWriteDescriptorSet(image_infos, *m_sampler, image_view, image.descriptor_sets[0], 1)); + + m_device.GetLogical().UpdateDescriptorSets(updates, {}); +} + +void FXAA::UploadImages(Scheduler& scheduler) { + if (m_images_ready) { + return; + } + + scheduler.Record([&](vk::CommandBuffer cmdbuf) { + for (auto& image : m_dynamic_images) { + ClearColorImage(cmdbuf, *image.image); + } + }); + scheduler.Finish(); + + m_images_ready = true; +} + +VkImageView FXAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, + VkImageView source_image_view) { + const Image& image{m_dynamic_images[image_index]}; + const VkImage output_image{*image.image}; + const VkDescriptorSet descriptor_set{image.descriptor_sets[0]}; + const VkFramebuffer framebuffer{*image.framebuffer}; + const VkRenderPass renderpass{*m_renderpass}; + const VkPipeline pipeline{*m_pipeline}; + const VkPipelineLayout layout{*m_pipeline_layout}; + const VkExtent2D extent{m_extent}; + + UploadImages(scheduler); + UpdateDescriptorSets(source_image_view, image_index); + + scheduler.RequestOutsideRenderPassOperationContext(); + scheduler.Record([=](vk::CommandBuffer cmdbuf) { + TransitionImageLayout(cmdbuf, source_image, VK_IMAGE_LAYOUT_GENERAL); + TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL); + BeginRenderPass(cmdbuf, renderpass, framebuffer, extent); + cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline); + cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, layout, 0, descriptor_set, {}); + cmdbuf.Draw(4, 1, 0, 0); + cmdbuf.EndRenderPass(); + TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL); + }); + + return *image.image_view; +} + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/fxaa.h b/src/video_core/renderer_vulkan/present/fxaa.h new file mode 100644 index 000000000..c083f3ff0 --- /dev/null +++ b/src/video_core/renderer_vulkan/present/fxaa.h @@ -0,0 +1,63 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include "video_core/renderer_vulkan/present/anti_alias_pass.h" +#include "video_core/vulkan_common/vulkan_memory_allocator.h" +#include "video_core/vulkan_common/vulkan_wrapper.h" + +namespace Vulkan { + +class Device; +class Scheduler; +class StagingBufferPool; + +class FXAA final : public AntiAliasPass { +public: + explicit FXAA(const Device& device, MemoryAllocator& allocator, size_t image_count, + VkExtent2D extent); + ~FXAA() override; + + VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, + VkImageView source_image_view) override; + +private: + void CreateImages(); + void CreateRenderPasses(); + void CreateSampler(); + void CreateShaders(); + void CreateDescriptorPool(); + void CreateDescriptorSetLayouts(); + void CreateDescriptorSets(); + void CreatePipelineLayouts(); + void CreatePipelines(); + void UpdateDescriptorSets(VkImageView image_view, size_t image_index); + void UploadImages(Scheduler& scheduler); + + const Device& m_device; + MemoryAllocator& m_allocator; + const VkExtent2D m_extent; + const u32 m_image_count; + + vk::ShaderModule m_vertex_shader{}; + vk::ShaderModule m_fragment_shader{}; + vk::DescriptorPool m_descriptor_pool{}; + vk::DescriptorSetLayout m_descriptor_set_layout{}; + vk::PipelineLayout m_pipeline_layout{}; + vk::Pipeline m_pipeline{}; + vk::RenderPass m_renderpass{}; + + struct Image { + vk::DescriptorSets descriptor_sets{}; + vk::Framebuffer framebuffer{}; + vk::Image image{}; + vk::ImageView image_view{}; + }; + std::vector m_dynamic_images{}; + bool m_images_ready{}; + + vk::Sampler m_sampler{}; +}; + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/smaa.cpp b/src/video_core/renderer_vulkan/present/smaa.cpp new file mode 100644 index 000000000..68cd22b08 --- /dev/null +++ b/src/video_core/renderer_vulkan/present/smaa.cpp @@ -0,0 +1,270 @@ +// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#include + +#include "common/assert.h" +#include "common/polyfill_ranges.h" + +#include "video_core/renderer_vulkan/present/smaa.h" +#include "video_core/renderer_vulkan/present/util.h" +#include "video_core/renderer_vulkan/vk_scheduler.h" +#include "video_core/renderer_vulkan/vk_shader_util.h" +#include "video_core/smaa_area_tex.h" +#include "video_core/smaa_search_tex.h" +#include "video_core/vulkan_common/vulkan_device.h" + +#include "video_core/host_shaders/smaa_blending_weight_calculation_frag_spv.h" +#include "video_core/host_shaders/smaa_blending_weight_calculation_vert_spv.h" +#include "video_core/host_shaders/smaa_edge_detection_frag_spv.h" +#include "video_core/host_shaders/smaa_edge_detection_vert_spv.h" +#include "video_core/host_shaders/smaa_neighborhood_blending_frag_spv.h" +#include "video_core/host_shaders/smaa_neighborhood_blending_vert_spv.h" + +namespace Vulkan { + +SMAA::SMAA(const Device& device, MemoryAllocator& allocator, size_t image_count, VkExtent2D extent) + : m_device(device), m_allocator(allocator), m_extent(extent), + m_image_count(static_cast(image_count)) { + CreateImages(); + CreateRenderPasses(); + CreateSampler(); + CreateShaders(); + CreateDescriptorPool(); + CreateDescriptorSetLayouts(); + CreateDescriptorSets(); + CreatePipelineLayouts(); + CreatePipelines(); +} + +SMAA::~SMAA() = default; + +void SMAA::CreateImages() { + static constexpr VkExtent2D area_extent{AREATEX_WIDTH, AREATEX_HEIGHT}; + static constexpr VkExtent2D search_extent{SEARCHTEX_WIDTH, SEARCHTEX_HEIGHT}; + + m_static_images[Area] = CreateWrappedImage(m_allocator, area_extent, VK_FORMAT_R8G8_UNORM); + m_static_images[Search] = CreateWrappedImage(m_allocator, search_extent, VK_FORMAT_R8_UNORM); + + m_static_image_views[Area] = + CreateWrappedImageView(m_device, m_static_images[Area], VK_FORMAT_R8G8_UNORM); + m_static_image_views[Search] = + CreateWrappedImageView(m_device, m_static_images[Search], VK_FORMAT_R8_UNORM); + + for (u32 i = 0; i < m_image_count; i++) { + Images& images = m_dynamic_images.emplace_back(); + + images.images[Blend] = + CreateWrappedImage(m_allocator, m_extent, VK_FORMAT_R16G16B16A16_SFLOAT); + images.images[Edges] = CreateWrappedImage(m_allocator, m_extent, VK_FORMAT_R16G16_SFLOAT); + images.images[Output] = + CreateWrappedImage(m_allocator, m_extent, VK_FORMAT_R16G16B16A16_SFLOAT); + + images.image_views[Blend] = + CreateWrappedImageView(m_device, images.images[Blend], VK_FORMAT_R16G16B16A16_SFLOAT); + images.image_views[Edges] = + CreateWrappedImageView(m_device, images.images[Edges], VK_FORMAT_R16G16_SFLOAT); + images.image_views[Output] = + CreateWrappedImageView(m_device, images.images[Output], VK_FORMAT_R16G16B16A16_SFLOAT); + } +} + +void SMAA::CreateRenderPasses() { + m_renderpasses[EdgeDetection] = CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16_SFLOAT); + m_renderpasses[BlendingWeightCalculation] = + CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16B16A16_SFLOAT); + m_renderpasses[NeighborhoodBlending] = + CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16B16A16_SFLOAT); + + for (auto& images : m_dynamic_images) { + images.framebuffers[EdgeDetection] = CreateWrappedFramebuffer( + m_device, m_renderpasses[EdgeDetection], images.image_views[Edges], m_extent); + + images.framebuffers[BlendingWeightCalculation] = + CreateWrappedFramebuffer(m_device, m_renderpasses[BlendingWeightCalculation], + images.image_views[Blend], m_extent); + + images.framebuffers[NeighborhoodBlending] = CreateWrappedFramebuffer( + m_device, m_renderpasses[NeighborhoodBlending], images.image_views[Output], m_extent); + } +} + +void SMAA::CreateSampler() { + m_sampler = CreateWrappedSampler(m_device); +} + +void SMAA::CreateShaders() { + // These match the order of the SMAAStage enum + static constexpr std::array vert_shader_sources{ + ARRAY_TO_SPAN(SMAA_EDGE_DETECTION_VERT_SPV), + ARRAY_TO_SPAN(SMAA_BLENDING_WEIGHT_CALCULATION_VERT_SPV), + ARRAY_TO_SPAN(SMAA_NEIGHBORHOOD_BLENDING_VERT_SPV), + }; + static constexpr std::array frag_shader_sources{ + ARRAY_TO_SPAN(SMAA_EDGE_DETECTION_FRAG_SPV), + ARRAY_TO_SPAN(SMAA_BLENDING_WEIGHT_CALCULATION_FRAG_SPV), + ARRAY_TO_SPAN(SMAA_NEIGHBORHOOD_BLENDING_FRAG_SPV), + }; + + for (size_t i = 0; i < MaxSMAAStage; i++) { + m_vertex_shaders[i] = CreateWrappedShaderModule(m_device, vert_shader_sources[i]); + m_fragment_shaders[i] = CreateWrappedShaderModule(m_device, frag_shader_sources[i]); + } +} + +void SMAA::CreateDescriptorPool() { + // Edge detection: 1 descriptor + // Blending weight calculation: 3 descriptors + // Neighborhood blending: 2 descriptors + + // 6 descriptors, 3 descriptor sets per image + m_descriptor_pool = CreateWrappedDescriptorPool(m_device, 6 * m_image_count, 3 * m_image_count); +} + +void SMAA::CreateDescriptorSetLayouts() { + m_descriptor_set_layouts[EdgeDetection] = CreateWrappedDescriptorSetLayout(m_device, 1); + m_descriptor_set_layouts[BlendingWeightCalculation] = + CreateWrappedDescriptorSetLayout(m_device, 3); + m_descriptor_set_layouts[NeighborhoodBlending] = CreateWrappedDescriptorSetLayout(m_device, 2); +} + +void SMAA::CreateDescriptorSets() { + std::vector layouts(m_descriptor_set_layouts.size()); + std::ranges::transform(m_descriptor_set_layouts, layouts.begin(), + [](auto& layout) { return *layout; }); + + for (auto& images : m_dynamic_images) { + images.descriptor_sets = CreateWrappedDescriptorSets(m_descriptor_pool, layouts); + } +} + +void SMAA::CreatePipelineLayouts() { + for (size_t i = 0; i < MaxSMAAStage; i++) { + m_pipeline_layouts[i] = CreateWrappedPipelineLayout(m_device, m_descriptor_set_layouts[i]); + } +} + +void SMAA::CreatePipelines() { + for (size_t i = 0; i < MaxSMAAStage; i++) { + m_pipelines[i] = + CreateWrappedPipeline(m_device, m_renderpasses[i], m_pipeline_layouts[i], + std::tie(m_vertex_shaders[i], m_fragment_shaders[i])); + } +} + +void SMAA::UpdateDescriptorSets(VkImageView image_view, size_t image_index) { + Images& images = m_dynamic_images[image_index]; + std::vector image_infos; + std::vector updates; + image_infos.reserve(6); + + updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, image_view, + images.descriptor_sets[EdgeDetection], 0)); + + updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, *images.image_views[Edges], + images.descriptor_sets[BlendingWeightCalculation], + 0)); + updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, *m_static_image_views[Area], + images.descriptor_sets[BlendingWeightCalculation], + 1)); + updates.push_back( + CreateWriteDescriptorSet(image_infos, *m_sampler, *m_static_image_views[Search], + images.descriptor_sets[BlendingWeightCalculation], 2)); + + updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, image_view, + images.descriptor_sets[NeighborhoodBlending], 0)); + updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, *images.image_views[Blend], + images.descriptor_sets[NeighborhoodBlending], 1)); + + m_device.GetLogical().UpdateDescriptorSets(updates, {}); +} + +void SMAA::UploadImages(Scheduler& scheduler) { + if (m_images_ready) { + return; + } + + static constexpr VkExtent2D area_extent{AREATEX_WIDTH, AREATEX_HEIGHT}; + static constexpr VkExtent2D search_extent{SEARCHTEX_WIDTH, SEARCHTEX_HEIGHT}; + + UploadImage(m_device, m_allocator, scheduler, m_static_images[Area], area_extent, + VK_FORMAT_R8G8_UNORM, ARRAY_TO_SPAN(areaTexBytes)); + UploadImage(m_device, m_allocator, scheduler, m_static_images[Search], search_extent, + VK_FORMAT_R8_UNORM, ARRAY_TO_SPAN(searchTexBytes)); + + scheduler.Record([&](vk::CommandBuffer cmdbuf) { + for (auto& images : m_dynamic_images) { + for (size_t i = 0; i < MaxDynamicImage; i++) { + ClearColorImage(cmdbuf, *images.images[i]); + } + } + }); + scheduler.Finish(); + + m_images_ready = true; +} + +VkImageView SMAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, + VkImageView source_image_view) { + Images& images = m_dynamic_images[image_index]; + + VkImage output_image = *images.images[Output]; + VkImage edges_image = *images.images[Edges]; + VkImage blend_image = *images.images[Blend]; + + VkDescriptorSet edge_detection_descriptor_set = images.descriptor_sets[EdgeDetection]; + VkDescriptorSet blending_weight_calculation_descriptor_set = + images.descriptor_sets[BlendingWeightCalculation]; + VkDescriptorSet neighborhood_blending_descriptor_set = + images.descriptor_sets[NeighborhoodBlending]; + + VkFramebuffer edge_detection_framebuffer = *images.framebuffers[EdgeDetection]; + VkFramebuffer blending_weight_calculation_framebuffer = + *images.framebuffers[BlendingWeightCalculation]; + VkFramebuffer neighborhood_blending_framebuffer = *images.framebuffers[NeighborhoodBlending]; + + UploadImages(scheduler); + UpdateDescriptorSets(source_image_view, image_index); + + scheduler.RequestOutsideRenderPassOperationContext(); + scheduler.Record([=, this](vk::CommandBuffer cmdbuf) { + TransitionImageLayout(cmdbuf, source_image, VK_IMAGE_LAYOUT_GENERAL); + TransitionImageLayout(cmdbuf, edges_image, VK_IMAGE_LAYOUT_GENERAL); + BeginRenderPass(cmdbuf, *m_renderpasses[EdgeDetection], edge_detection_framebuffer, + m_extent); + cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *m_pipelines[EdgeDetection]); + cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, + *m_pipeline_layouts[EdgeDetection], 0, + edge_detection_descriptor_set, {}); + cmdbuf.Draw(3, 1, 0, 0); + cmdbuf.EndRenderPass(); + + TransitionImageLayout(cmdbuf, edges_image, VK_IMAGE_LAYOUT_GENERAL); + TransitionImageLayout(cmdbuf, blend_image, VK_IMAGE_LAYOUT_GENERAL); + BeginRenderPass(cmdbuf, *m_renderpasses[BlendingWeightCalculation], + blending_weight_calculation_framebuffer, m_extent); + cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, + *m_pipelines[BlendingWeightCalculation]); + cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, + *m_pipeline_layouts[BlendingWeightCalculation], 0, + blending_weight_calculation_descriptor_set, {}); + cmdbuf.Draw(3, 1, 0, 0); + cmdbuf.EndRenderPass(); + + TransitionImageLayout(cmdbuf, blend_image, VK_IMAGE_LAYOUT_GENERAL); + TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL); + BeginRenderPass(cmdbuf, *m_renderpasses[NeighborhoodBlending], + neighborhood_blending_framebuffer, m_extent); + cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *m_pipelines[NeighborhoodBlending]); + cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, + *m_pipeline_layouts[NeighborhoodBlending], 0, + neighborhood_blending_descriptor_set, {}); + cmdbuf.Draw(3, 1, 0, 0); + cmdbuf.EndRenderPass(); + TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL); + }); + + return *images.image_views[Output]; +} + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/smaa.h b/src/video_core/renderer_vulkan/present/smaa.h new file mode 100644 index 000000000..3d6707d48 --- /dev/null +++ b/src/video_core/renderer_vulkan/present/smaa.h @@ -0,0 +1,87 @@ +// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include +#include "video_core/renderer_vulkan/present/anti_alias_pass.h" +#include "video_core/vulkan_common/vulkan_memory_allocator.h" +#include "video_core/vulkan_common/vulkan_wrapper.h" + +namespace Vulkan { + +class Device; +class Scheduler; +class StagingBufferPool; + +class SMAA final : public AntiAliasPass { +public: + explicit SMAA(const Device& device, MemoryAllocator& allocator, size_t image_count, + VkExtent2D extent); + ~SMAA() override; + + VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, + VkImageView source_image_view) override; + +private: + enum SMAAStage { + EdgeDetection = 0, + BlendingWeightCalculation = 1, + NeighborhoodBlending = 2, + MaxSMAAStage = 3, + }; + + enum StaticImageType { + Area = 0, + Search = 1, + MaxStaticImage = 2, + }; + + enum DynamicImageType { + Blend = 0, + Edges = 1, + Output = 2, + MaxDynamicImage = 3, + }; + + void CreateImages(); + void CreateRenderPasses(); + void CreateSampler(); + void CreateShaders(); + void CreateDescriptorPool(); + void CreateDescriptorSetLayouts(); + void CreateDescriptorSets(); + void CreatePipelineLayouts(); + void CreatePipelines(); + void UpdateDescriptorSets(VkImageView image_view, size_t image_index); + void UploadImages(Scheduler& scheduler); + + const Device& m_device; + MemoryAllocator& m_allocator; + const VkExtent2D m_extent; + const u32 m_image_count; + + vk::DescriptorPool m_descriptor_pool{}; + std::array m_descriptor_set_layouts{}; + std::array m_pipeline_layouts{}; + std::array m_vertex_shaders{}; + std::array m_fragment_shaders{}; + std::array m_pipelines{}; + std::array m_renderpasses{}; + + std::array m_static_images{}; + std::array m_static_image_views{}; + + struct Images { + vk::DescriptorSets descriptor_sets{}; + std::array images{}; + std::array image_views{}; + std::array framebuffers{}; + }; + std::vector m_dynamic_images{}; + bool m_images_ready{}; + + vk::Sampler m_sampler{}; +}; + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/util.cpp b/src/video_core/renderer_vulkan/present/util.cpp new file mode 100644 index 000000000..a445b213e --- /dev/null +++ b/src/video_core/renderer_vulkan/present/util.cpp @@ -0,0 +1,488 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#include "common/assert.h" +#include "common/polyfill_ranges.h" +#include "video_core/renderer_vulkan/present/util.h" + +namespace Vulkan { + +vk::Image CreateWrappedImage(MemoryAllocator& allocator, VkExtent2D dimensions, VkFormat format) { + const VkImageCreateInfo image_ci{ + .sType = VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .imageType = VK_IMAGE_TYPE_2D, + .format = format, + .extent = {.width = dimensions.width, .height = dimensions.height, .depth = 1}, + .mipLevels = 1, + .arrayLayers = 1, + .samples = VK_SAMPLE_COUNT_1_BIT, + .tiling = VK_IMAGE_TILING_OPTIMAL, + .usage = VK_IMAGE_USAGE_TRANSFER_DST_BIT | VK_IMAGE_USAGE_STORAGE_BIT | + VK_IMAGE_USAGE_SAMPLED_BIT | VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT, + .sharingMode = VK_SHARING_MODE_EXCLUSIVE, + .queueFamilyIndexCount = 0, + .pQueueFamilyIndices = nullptr, + .initialLayout = VK_IMAGE_LAYOUT_UNDEFINED, + }; + return allocator.CreateImage(image_ci); +} + +void TransitionImageLayout(vk::CommandBuffer& cmdbuf, VkImage image, VkImageLayout target_layout, + VkImageLayout source_layout) { + constexpr VkFlags flags{VK_ACCESS_COLOR_ATTACHMENT_READ_BIT | + VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT | VK_ACCESS_SHADER_READ_BIT}; + const VkImageMemoryBarrier barrier{ + .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER, + .pNext = nullptr, + .srcAccessMask = flags, + .dstAccessMask = flags, + .oldLayout = source_layout, + .newLayout = target_layout, + .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, + .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, + .image = image, + .subresourceRange{ + .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .baseMipLevel = 0, + .levelCount = 1, + .baseArrayLayer = 0, + .layerCount = 1, + }, + }; + cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, + 0, barrier); +} + +void UploadImage(const Device& device, MemoryAllocator& allocator, Scheduler& scheduler, + vk::Image& image, VkExtent2D dimensions, VkFormat format, + std::span initial_contents) { + const VkBufferCreateInfo upload_ci = { + .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .size = initial_contents.size_bytes(), + .usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT, + .sharingMode = VK_SHARING_MODE_EXCLUSIVE, + .queueFamilyIndexCount = 0, + .pQueueFamilyIndices = nullptr, + }; + auto upload_buffer = allocator.CreateBuffer(upload_ci, MemoryUsage::Upload); + std::ranges::copy(initial_contents, upload_buffer.Mapped().begin()); + upload_buffer.Flush(); + + const std::array regions{{{ + .bufferOffset = 0, + .bufferRowLength = dimensions.width, + .bufferImageHeight = dimensions.height, + .imageSubresource{.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .mipLevel = 0, + .baseArrayLayer = 0, + .layerCount = 1}, + .imageOffset{}, + .imageExtent{.width = dimensions.width, .height = dimensions.height, .depth = 1}, + }}}; + + scheduler.RequestOutsideRenderPassOperationContext(); + scheduler.Record([&](vk::CommandBuffer cmdbuf) { + TransitionImageLayout(cmdbuf, *image, VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, + VK_IMAGE_LAYOUT_UNDEFINED); + cmdbuf.CopyBufferToImage(*upload_buffer, *image, VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, + regions); + TransitionImageLayout(cmdbuf, *image, VK_IMAGE_LAYOUT_GENERAL, + VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL); + }); + scheduler.Finish(); +} + +vk::ImageView CreateWrappedImageView(const Device& device, vk::Image& image, VkFormat format) { + return device.GetLogical().CreateImageView(VkImageViewCreateInfo{ + .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .image = *image, + .viewType = VK_IMAGE_VIEW_TYPE_2D, + .format = format, + .components{}, + .subresourceRange{.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .baseMipLevel = 0, + .levelCount = 1, + .baseArrayLayer = 0, + .layerCount = 1}, + }); +} + +vk::RenderPass CreateWrappedRenderPass(const Device& device, VkFormat format) { + const VkAttachmentDescription attachment{ + .flags = VK_ATTACHMENT_DESCRIPTION_MAY_ALIAS_BIT, + .format = format, + .samples = VK_SAMPLE_COUNT_1_BIT, + .loadOp = VK_ATTACHMENT_LOAD_OP_LOAD, + .storeOp = VK_ATTACHMENT_STORE_OP_STORE, + .stencilLoadOp = VK_ATTACHMENT_LOAD_OP_LOAD, + .stencilStoreOp = VK_ATTACHMENT_STORE_OP_STORE, + .initialLayout = VK_IMAGE_LAYOUT_GENERAL, + .finalLayout = VK_IMAGE_LAYOUT_GENERAL, + }; + + constexpr VkAttachmentReference color_attachment_ref{ + .attachment = 0, + .layout = VK_IMAGE_LAYOUT_GENERAL, + }; + + const VkSubpassDescription subpass_description{ + .flags = 0, + .pipelineBindPoint = VK_PIPELINE_BIND_POINT_GRAPHICS, + .inputAttachmentCount = 0, + .pInputAttachments = nullptr, + .colorAttachmentCount = 1, + .pColorAttachments = &color_attachment_ref, + .pResolveAttachments = nullptr, + .pDepthStencilAttachment = nullptr, + .preserveAttachmentCount = 0, + .pPreserveAttachments = nullptr, + }; + + constexpr VkSubpassDependency dependency{ + .srcSubpass = VK_SUBPASS_EXTERNAL, + .dstSubpass = 0, + .srcStageMask = VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT, + .dstStageMask = VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT, + .srcAccessMask = 0, + .dstAccessMask = VK_ACCESS_COLOR_ATTACHMENT_READ_BIT | VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT, + .dependencyFlags = 0, + }; + + return device.GetLogical().CreateRenderPass(VkRenderPassCreateInfo{ + .sType = VK_STRUCTURE_TYPE_RENDER_PASS_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .attachmentCount = 1, + .pAttachments = &attachment, + .subpassCount = 1, + .pSubpasses = &subpass_description, + .dependencyCount = 1, + .pDependencies = &dependency, + }); +} + +vk::Framebuffer CreateWrappedFramebuffer(const Device& device, vk::RenderPass& render_pass, + vk::ImageView& dest_image, VkExtent2D extent) { + return device.GetLogical().CreateFramebuffer(VkFramebufferCreateInfo{ + .sType = VK_STRUCTURE_TYPE_FRAMEBUFFER_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .renderPass = *render_pass, + .attachmentCount = 1, + .pAttachments = dest_image.address(), + .width = extent.width, + .height = extent.height, + .layers = 1, + }); +} + +vk::Sampler CreateWrappedSampler(const Device& device, VkFilter filter) { + return device.GetLogical().CreateSampler(VkSamplerCreateInfo{ + .sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .magFilter = filter, + .minFilter = filter, + .mipmapMode = VK_SAMPLER_MIPMAP_MODE_LINEAR, + .addressModeU = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE, + .addressModeV = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE, + .addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE, + .mipLodBias = 0.0f, + .anisotropyEnable = VK_FALSE, + .maxAnisotropy = 0.0f, + .compareEnable = VK_FALSE, + .compareOp = VK_COMPARE_OP_NEVER, + .minLod = 0.0f, + .maxLod = 0.0f, + .borderColor = VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK, + .unnormalizedCoordinates = VK_FALSE, + }); +} + +vk::ShaderModule CreateWrappedShaderModule(const Device& device, std::span code) { + return device.GetLogical().CreateShaderModule(VkShaderModuleCreateInfo{ + .sType = VK_STRUCTURE_TYPE_SHADER_MODULE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .codeSize = code.size_bytes(), + .pCode = code.data(), + }); +} + +vk::DescriptorPool CreateWrappedDescriptorPool(const Device& device, u32 max_descriptors, + u32 max_sets) { + const VkDescriptorPoolSize pool_size{ + .type = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + .descriptorCount = static_cast(max_descriptors), + }; + + return device.GetLogical().CreateDescriptorPool(VkDescriptorPoolCreateInfo{ + .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .maxSets = max_sets, + .poolSizeCount = 1, + .pPoolSizes = &pool_size, + }); +} + +vk::DescriptorSetLayout CreateWrappedDescriptorSetLayout(const Device& device, + u32 max_sampler_bindings) { + std::vector bindings(max_sampler_bindings); + for (u32 i = 0; i < max_sampler_bindings; i++) { + bindings[i] = { + .binding = i, + .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + .descriptorCount = 1, + .stageFlags = VK_SHADER_STAGE_VERTEX_BIT | VK_SHADER_STAGE_FRAGMENT_BIT, + .pImmutableSamplers = nullptr, + }; + } + + return device.GetLogical().CreateDescriptorSetLayout(VkDescriptorSetLayoutCreateInfo{ + .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .bindingCount = static_cast(bindings.size()), + .pBindings = bindings.data(), + }); +} + +vk::DescriptorSets CreateWrappedDescriptorSets(vk::DescriptorPool& pool, + vk::Span layouts) { + return pool.Allocate(VkDescriptorSetAllocateInfo{ + .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_ALLOCATE_INFO, + .pNext = nullptr, + .descriptorPool = *pool, + .descriptorSetCount = layouts.size(), + .pSetLayouts = layouts.data(), + }); +} + +vk::PipelineLayout CreateWrappedPipelineLayout(const Device& device, + vk::DescriptorSetLayout& layout) { + return device.GetLogical().CreatePipelineLayout(VkPipelineLayoutCreateInfo{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .setLayoutCount = 1, + .pSetLayouts = layout.address(), + .pushConstantRangeCount = 0, + .pPushConstantRanges = nullptr, + }); +} + +vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderpass, + vk::PipelineLayout& layout, + std::tuple shaders) { + const std::array shader_stages{{ + { + .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .stage = VK_SHADER_STAGE_VERTEX_BIT, + .module = *std::get<0>(shaders), + .pName = "main", + .pSpecializationInfo = nullptr, + }, + { + .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .stage = VK_SHADER_STAGE_FRAGMENT_BIT, + .module = *std::get<1>(shaders), + .pName = "main", + .pSpecializationInfo = nullptr, + }, + }}; + + constexpr VkPipelineVertexInputStateCreateInfo vertex_input_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .vertexBindingDescriptionCount = 0, + .pVertexBindingDescriptions = nullptr, + .vertexAttributeDescriptionCount = 0, + .pVertexAttributeDescriptions = nullptr, + }; + + constexpr VkPipelineInputAssemblyStateCreateInfo input_assembly_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .topology = VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP, + .primitiveRestartEnable = VK_FALSE, + }; + + constexpr VkPipelineViewportStateCreateInfo viewport_state_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .viewportCount = 1, + .pViewports = nullptr, + .scissorCount = 1, + .pScissors = nullptr, + }; + + constexpr VkPipelineRasterizationStateCreateInfo rasterization_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .depthClampEnable = VK_FALSE, + .rasterizerDiscardEnable = VK_FALSE, + .polygonMode = VK_POLYGON_MODE_FILL, + .cullMode = VK_CULL_MODE_NONE, + .frontFace = VK_FRONT_FACE_CLOCKWISE, + .depthBiasEnable = VK_FALSE, + .depthBiasConstantFactor = 0.0f, + .depthBiasClamp = 0.0f, + .depthBiasSlopeFactor = 0.0f, + .lineWidth = 1.0f, + }; + + constexpr VkPipelineMultisampleStateCreateInfo multisampling_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_MULTISAMPLE_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .rasterizationSamples = VK_SAMPLE_COUNT_1_BIT, + .sampleShadingEnable = VK_FALSE, + .minSampleShading = 0.0f, + .pSampleMask = nullptr, + .alphaToCoverageEnable = VK_FALSE, + .alphaToOneEnable = VK_FALSE, + }; + + constexpr VkPipelineColorBlendAttachmentState color_blend_attachment{ + .blendEnable = VK_FALSE, + .srcColorBlendFactor = VK_BLEND_FACTOR_ZERO, + .dstColorBlendFactor = VK_BLEND_FACTOR_ZERO, + .colorBlendOp = VK_BLEND_OP_ADD, + .srcAlphaBlendFactor = VK_BLEND_FACTOR_ZERO, + .dstAlphaBlendFactor = VK_BLEND_FACTOR_ZERO, + .alphaBlendOp = VK_BLEND_OP_ADD, + .colorWriteMask = VK_COLOR_COMPONENT_R_BIT | VK_COLOR_COMPONENT_G_BIT | + VK_COLOR_COMPONENT_B_BIT | VK_COLOR_COMPONENT_A_BIT, + }; + + const VkPipelineColorBlendStateCreateInfo color_blend_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_COLOR_BLEND_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .logicOpEnable = VK_FALSE, + .logicOp = VK_LOGIC_OP_COPY, + .attachmentCount = 1, + .pAttachments = &color_blend_attachment, + .blendConstants = {0.0f, 0.0f, 0.0f, 0.0f}, + }; + + constexpr std::array dynamic_states{ + VK_DYNAMIC_STATE_VIEWPORT, + VK_DYNAMIC_STATE_SCISSOR, + }; + + const VkPipelineDynamicStateCreateInfo dynamic_state_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_DYNAMIC_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .dynamicStateCount = static_cast(dynamic_states.size()), + .pDynamicStates = dynamic_states.data(), + }; + + return device.GetLogical().CreateGraphicsPipeline(VkGraphicsPipelineCreateInfo{ + .sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .stageCount = static_cast(shader_stages.size()), + .pStages = shader_stages.data(), + .pVertexInputState = &vertex_input_ci, + .pInputAssemblyState = &input_assembly_ci, + .pTessellationState = nullptr, + .pViewportState = &viewport_state_ci, + .pRasterizationState = &rasterization_ci, + .pMultisampleState = &multisampling_ci, + .pDepthStencilState = nullptr, + .pColorBlendState = &color_blend_ci, + .pDynamicState = &dynamic_state_ci, + .layout = *layout, + .renderPass = *renderpass, + .subpass = 0, + .basePipelineHandle = 0, + .basePipelineIndex = 0, + }); +} + +VkWriteDescriptorSet CreateWriteDescriptorSet(std::vector& images, + VkSampler sampler, VkImageView view, + VkDescriptorSet set, u32 binding) { + ASSERT(images.capacity() > images.size()); + auto& image_info = images.emplace_back(VkDescriptorImageInfo{ + .sampler = sampler, + .imageView = view, + .imageLayout = VK_IMAGE_LAYOUT_GENERAL, + }); + + return VkWriteDescriptorSet{ + .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET, + .pNext = nullptr, + .dstSet = set, + .dstBinding = binding, + .dstArrayElement = 0, + .descriptorCount = 1, + .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + .pImageInfo = &image_info, + .pBufferInfo = nullptr, + .pTexelBufferView = nullptr, + }; +} + +void ClearColorImage(vk::CommandBuffer& cmdbuf, VkImage image) { + static constexpr std::array subresources{{{ + .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .baseMipLevel = 0, + .levelCount = 1, + .baseArrayLayer = 0, + .layerCount = 1, + }}}; + TransitionImageLayout(cmdbuf, image, VK_IMAGE_LAYOUT_GENERAL, VK_IMAGE_LAYOUT_UNDEFINED); + cmdbuf.ClearColorImage(image, VK_IMAGE_LAYOUT_GENERAL, {}, subresources); +} + +void BeginRenderPass(vk::CommandBuffer& cmdbuf, VkRenderPass render_pass, VkFramebuffer framebuffer, + VkExtent2D extent) { + const VkRenderPassBeginInfo renderpass_bi{ + .sType = VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO, + .pNext = nullptr, + .renderPass = render_pass, + .framebuffer = framebuffer, + .renderArea{ + .offset{}, + .extent = extent, + }, + .clearValueCount = 0, + .pClearValues = nullptr, + }; + cmdbuf.BeginRenderPass(renderpass_bi, VK_SUBPASS_CONTENTS_INLINE); + + const VkViewport viewport{ + .x = 0.0f, + .y = 0.0f, + .width = static_cast(extent.width), + .height = static_cast(extent.height), + .minDepth = 0.0f, + .maxDepth = 1.0f, + }; + const VkRect2D scissor{ + .offset = {0, 0}, + .extent = extent, + }; + cmdbuf.SetViewport(0, viewport); + cmdbuf.SetScissor(0, scissor); +} + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/util.h b/src/video_core/renderer_vulkan/present/util.h new file mode 100644 index 000000000..93cfdd16b --- /dev/null +++ b/src/video_core/renderer_vulkan/present/util.h @@ -0,0 +1,46 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include "video_core/renderer_vulkan/vk_scheduler.h" +#include "video_core/vulkan_common/vulkan_memory_allocator.h" +#include "video_core/vulkan_common/vulkan_wrapper.h" + +namespace Vulkan { + +#define ARRAY_TO_SPAN(a) std::span(a, (sizeof(a) / sizeof(a[0]))) + +vk::Image CreateWrappedImage(MemoryAllocator& allocator, VkExtent2D dimensions, VkFormat format); +void TransitionImageLayout(vk::CommandBuffer& cmdbuf, VkImage image, VkImageLayout target_layout, + VkImageLayout source_layout = VK_IMAGE_LAYOUT_GENERAL); +void UploadImage(const Device& device, MemoryAllocator& allocator, Scheduler& scheduler, + vk::Image& image, VkExtent2D dimensions, VkFormat format, + std::span initial_contents = {}); +void ClearColorImage(vk::CommandBuffer& cmdbuf, VkImage image); + +vk::ImageView CreateWrappedImageView(const Device& device, vk::Image& image, VkFormat format); +vk::RenderPass CreateWrappedRenderPass(const Device& device, VkFormat format); +vk::Framebuffer CreateWrappedFramebuffer(const Device& device, vk::RenderPass& render_pass, + vk::ImageView& dest_image, VkExtent2D extent); +vk::Sampler CreateWrappedSampler(const Device& device, VkFilter filter = VK_FILTER_LINEAR); +vk::ShaderModule CreateWrappedShaderModule(const Device& device, std::span code); +vk::DescriptorPool CreateWrappedDescriptorPool(const Device& device, u32 max_sampler_bindings, + u32 max_sets); +vk::DescriptorSetLayout CreateWrappedDescriptorSetLayout(const Device& device, + u32 max_sampler_bindings); +vk::DescriptorSets CreateWrappedDescriptorSets(vk::DescriptorPool& pool, + vk::Span layouts); +vk::PipelineLayout CreateWrappedPipelineLayout(const Device& device, + vk::DescriptorSetLayout& layout); +vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderpass, + vk::PipelineLayout& layout, + std::tuple shaders); +VkWriteDescriptorSet CreateWriteDescriptorSet(std::vector& images, + VkSampler sampler, VkImageView view, + VkDescriptorSet set, u32 binding); + +void BeginRenderPass(vk::CommandBuffer& cmdbuf, VkRenderPass render_pass, VkFramebuffer framebuffer, + VkExtent2D extent); + +} // namespace Vulkan -- cgit v1.2.3 From 0c2e5b64c9fb985a40e5afec898d1f370cbad23e Mon Sep 17 00:00:00 2001 From: Liam Date: Sun, 14 Jan 2024 01:46:19 -0500 Subject: renderer_vulkan: split up blit screen resources into separate antialias and window adapt passes --- src/video_core/renderer_vulkan/present/filters.cpp | 70 +++ src/video_core/renderer_vulkan/present/filters.h | 30 ++ src/video_core/renderer_vulkan/present/util.cpp | 50 ++ src/video_core/renderer_vulkan/present/util.h | 2 + .../renderer_vulkan/present/window_adapt_pass.cpp | 512 +++++++++++++++++++++ .../renderer_vulkan/present/window_adapt_pass.h | 71 +++ 6 files changed, 735 insertions(+) create mode 100644 src/video_core/renderer_vulkan/present/filters.cpp create mode 100644 src/video_core/renderer_vulkan/present/filters.h create mode 100644 src/video_core/renderer_vulkan/present/window_adapt_pass.cpp create mode 100644 src/video_core/renderer_vulkan/present/window_adapt_pass.h (limited to 'src/video_core/renderer_vulkan/present') diff --git a/src/video_core/renderer_vulkan/present/filters.cpp b/src/video_core/renderer_vulkan/present/filters.cpp new file mode 100644 index 000000000..ee6239cc4 --- /dev/null +++ b/src/video_core/renderer_vulkan/present/filters.cpp @@ -0,0 +1,70 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#include "common/common_types.h" + +#include "video_core/host_shaders/present_bicubic_frag_spv.h" +#include "video_core/host_shaders/present_gaussian_frag_spv.h" +#include "video_core/host_shaders/vulkan_present_frag_spv.h" +#include "video_core/host_shaders/vulkan_present_scaleforce_fp16_frag_spv.h" +#include "video_core/host_shaders/vulkan_present_scaleforce_fp32_frag_spv.h" +#include "video_core/renderer_vulkan/present/filters.h" +#include "video_core/renderer_vulkan/present/util.h" +#include "video_core/renderer_vulkan/vk_shader_util.h" +#include "video_core/vulkan_common/vulkan_device.h" + +namespace Vulkan { + +namespace { + +vk::ShaderModule SelectScaleForceShader(const Device& device) { + if (device.IsFloat16Supported()) { + return BuildShader(device, VULKAN_PRESENT_SCALEFORCE_FP16_FRAG_SPV); + } else { + return BuildShader(device, VULKAN_PRESENT_SCALEFORCE_FP32_FRAG_SPV); + } +} + +} // Anonymous namespace + +std::unique_ptr MakeNearestNeighbor(const Device& device, + const MemoryAllocator& memory_allocator, + size_t image_count, VkFormat frame_format) { + return std::make_unique(device, memory_allocator, image_count, frame_format, + CreateNearestNeighborSampler(device), + BuildShader(device, VULKAN_PRESENT_FRAG_SPV)); +} + +std::unique_ptr MakeBilinear(const Device& device, + const MemoryAllocator& memory_allocator, + size_t image_count, VkFormat frame_format) { + return std::make_unique(device, memory_allocator, image_count, frame_format, + CreateBilinearSampler(device), + BuildShader(device, VULKAN_PRESENT_FRAG_SPV)); +} + +std::unique_ptr MakeBicubic(const Device& device, + const MemoryAllocator& memory_allocator, + size_t image_count, VkFormat frame_format) { + return std::make_unique(device, memory_allocator, image_count, frame_format, + CreateBilinearSampler(device), + BuildShader(device, PRESENT_BICUBIC_FRAG_SPV)); +} + +std::unique_ptr MakeGaussian(const Device& device, + const MemoryAllocator& memory_allocator, + size_t image_count, VkFormat frame_format) { + return std::make_unique(device, memory_allocator, image_count, frame_format, + CreateBilinearSampler(device), + BuildShader(device, PRESENT_GAUSSIAN_FRAG_SPV)); +} + +std::unique_ptr MakeScaleForce(const Device& device, + const MemoryAllocator& memory_allocator, + size_t image_count, VkFormat frame_format) { + return std::make_unique(device, memory_allocator, image_count, frame_format, + CreateBilinearSampler(device), + SelectScaleForceShader(device)); +} + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/filters.h b/src/video_core/renderer_vulkan/present/filters.h new file mode 100644 index 000000000..42d7052da --- /dev/null +++ b/src/video_core/renderer_vulkan/present/filters.h @@ -0,0 +1,30 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include "video_core/renderer_vulkan/present/window_adapt_pass.h" + +namespace Vulkan { + +std::unique_ptr MakeNearestNeighbor(const Device& device, + const MemoryAllocator& memory_allocator, + size_t image_count, VkFormat frame_format); + +std::unique_ptr MakeBilinear(const Device& device, + const MemoryAllocator& memory_allocator, + size_t image_count, VkFormat frame_format); + +std::unique_ptr MakeBicubic(const Device& device, + const MemoryAllocator& memory_allocator, + size_t image_count, VkFormat frame_format); + +std::unique_ptr MakeGaussian(const Device& device, + const MemoryAllocator& memory_allocator, + size_t image_count, VkFormat frame_format); + +std::unique_ptr MakeScaleForce(const Device& device, + const MemoryAllocator& memory_allocator, + size_t image_count, VkFormat frame_format); + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/util.cpp b/src/video_core/renderer_vulkan/present/util.cpp index a445b213e..cd6061101 100644 --- a/src/video_core/renderer_vulkan/present/util.cpp +++ b/src/video_core/renderer_vulkan/present/util.cpp @@ -441,6 +441,56 @@ VkWriteDescriptorSet CreateWriteDescriptorSet(std::vector }; } +vk::Sampler CreateBilinearSampler(const Device& device) { + const VkSamplerCreateInfo ci{ + .sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .magFilter = VK_FILTER_LINEAR, + .minFilter = VK_FILTER_LINEAR, + .mipmapMode = VK_SAMPLER_MIPMAP_MODE_NEAREST, + .addressModeU = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER, + .addressModeV = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER, + .addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER, + .mipLodBias = 0.0f, + .anisotropyEnable = VK_FALSE, + .maxAnisotropy = 0.0f, + .compareEnable = VK_FALSE, + .compareOp = VK_COMPARE_OP_NEVER, + .minLod = 0.0f, + .maxLod = 0.0f, + .borderColor = VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK, + .unnormalizedCoordinates = VK_FALSE, + }; + + return device.GetLogical().CreateSampler(ci); +} + +vk::Sampler CreateNearestNeighborSampler(const Device& device) { + const VkSamplerCreateInfo ci_nn{ + .sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .magFilter = VK_FILTER_NEAREST, + .minFilter = VK_FILTER_NEAREST, + .mipmapMode = VK_SAMPLER_MIPMAP_MODE_NEAREST, + .addressModeU = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER, + .addressModeV = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER, + .addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER, + .mipLodBias = 0.0f, + .anisotropyEnable = VK_FALSE, + .maxAnisotropy = 0.0f, + .compareEnable = VK_FALSE, + .compareOp = VK_COMPARE_OP_NEVER, + .minLod = 0.0f, + .maxLod = 0.0f, + .borderColor = VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK, + .unnormalizedCoordinates = VK_FALSE, + }; + + return device.GetLogical().CreateSampler(ci_nn); +} + void ClearColorImage(vk::CommandBuffer& cmdbuf, VkImage image) { static constexpr std::array subresources{{{ .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, diff --git a/src/video_core/renderer_vulkan/present/util.h b/src/video_core/renderer_vulkan/present/util.h index 93cfdd16b..ea9a26c3d 100644 --- a/src/video_core/renderer_vulkan/present/util.h +++ b/src/video_core/renderer_vulkan/present/util.h @@ -39,6 +39,8 @@ vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderp VkWriteDescriptorSet CreateWriteDescriptorSet(std::vector& images, VkSampler sampler, VkImageView view, VkDescriptorSet set, u32 binding); +vk::Sampler CreateBilinearSampler(const Device& device); +vk::Sampler CreateNearestNeighborSampler(const Device& device); void BeginRenderPass(vk::CommandBuffer& cmdbuf, VkRenderPass render_pass, VkFramebuffer framebuffer, VkExtent2D extent); diff --git a/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp b/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp new file mode 100644 index 000000000..7fd9ecd22 --- /dev/null +++ b/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp @@ -0,0 +1,512 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#include + +#include "core/frontend/framebuffer_layout.h" +#include "video_core/host_shaders/vulkan_present_vert_spv.h" +#include "video_core/renderer_vulkan/present/util.h" +#include "video_core/renderer_vulkan/present/window_adapt_pass.h" +#include "video_core/renderer_vulkan/vk_present_manager.h" +#include "video_core/renderer_vulkan/vk_shader_util.h" +#include "video_core/vulkan_common/vulkan_device.h" +#include "video_core/vulkan_common/vulkan_memory_allocator.h" + +namespace Vulkan { + +namespace { + +struct ScreenRectVertex { + ScreenRectVertex() = default; + explicit ScreenRectVertex(f32 x, f32 y, f32 u, f32 v) : position{{x, y}}, tex_coord{{u, v}} {} + + std::array position; + std::array tex_coord; + + static VkVertexInputBindingDescription GetDescription() { + return { + .binding = 0, + .stride = sizeof(ScreenRectVertex), + .inputRate = VK_VERTEX_INPUT_RATE_VERTEX, + }; + } + + static std::array GetAttributes() { + return {{ + { + .location = 0, + .binding = 0, + .format = VK_FORMAT_R32G32_SFLOAT, + .offset = offsetof(ScreenRectVertex, position), + }, + { + .location = 1, + .binding = 0, + .format = VK_FORMAT_R32G32_SFLOAT, + .offset = offsetof(ScreenRectVertex, tex_coord), + }, + }}; + } +}; + +std::array MakeOrthographicMatrix(f32 width, f32 height) { + // clang-format off + return { 2.f / width, 0.f, 0.f, 0.f, + 0.f, 2.f / height, 0.f, 0.f, + 0.f, 0.f, 1.f, 0.f, + -1.f, -1.f, 0.f, 1.f}; + // clang-format on +} + +} // Anonymous namespace + +struct WindowAdaptPass::BufferData { + struct { + std::array modelview_matrix; + } uniform; + + std::array vertices; +}; + +WindowAdaptPass::WindowAdaptPass(const Device& device_, const MemoryAllocator& memory_allocator, + size_t num_images, VkFormat frame_format, vk::Sampler&& sampler_, + vk::ShaderModule&& fragment_shader_) + : device(device_), sampler(std::move(sampler_)), fragment_shader(std::move(fragment_shader_)) { + CreateDescriptorPool(num_images); + CreateDescriptorSetLayout(); + CreateDescriptorSets(num_images); + CreatePipelineLayout(); + CreateVertexShader(); + CreateRenderPass(frame_format); + CreatePipeline(); + CreateBuffer(memory_allocator); +} + +WindowAdaptPass::~WindowAdaptPass() = default; + +void WindowAdaptPass::Draw(Scheduler& scheduler, size_t image_index, VkImageView src_image_view, + VkExtent2D src_image_extent, const Common::Rectangle& crop_rect, + const Layout::FramebufferLayout& layout, Frame* dst) { + ConfigureLayout(image_index, src_image_view, layout, crop_rect); + + const VkFramebuffer host_framebuffer{*dst->framebuffer}; + const VkRenderPass renderpass{*render_pass}; + const VkPipeline graphics_pipeline{*pipeline}; + const VkDescriptorSet descriptor_set{descriptor_sets[image_index]}; + const VkExtent2D render_area{ + .width = dst->width, + .height = dst->height, + }; + + scheduler.Record([=](vk::CommandBuffer cmdbuf) { + const f32 bg_red = Settings::values.bg_red.GetValue() / 255.0f; + const f32 bg_green = Settings::values.bg_green.GetValue() / 255.0f; + const f32 bg_blue = Settings::values.bg_blue.GetValue() / 255.0f; + const VkClearValue clear_color{ + .color = {.float32 = {bg_red, bg_green, bg_blue, 1.0f}}, + }; + const VkRenderPassBeginInfo renderpass_bi{ + .sType = VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO, + .pNext = nullptr, + .renderPass = renderpass, + .framebuffer = host_framebuffer, + .renderArea = + { + .offset = {0, 0}, + .extent = render_area, + }, + .clearValueCount = 1, + .pClearValues = &clear_color, + }; + const VkViewport viewport{ + .x = 0.0f, + .y = 0.0f, + .width = static_cast(render_area.width), + .height = static_cast(render_area.height), + .minDepth = 0.0f, + .maxDepth = 1.0f, + }; + const VkRect2D scissor{ + .offset = {0, 0}, + .extent = render_area, + }; + cmdbuf.BeginRenderPass(renderpass_bi, VK_SUBPASS_CONTENTS_INLINE); + cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, graphics_pipeline); + cmdbuf.SetViewport(0, viewport); + cmdbuf.SetScissor(0, scissor); + cmdbuf.BindVertexBuffer(0, *buffer, offsetof(BufferData, vertices)); + cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, *pipeline_layout, 0, + descriptor_set, {}); + cmdbuf.Draw(4, 1, 0, 0); + cmdbuf.EndRenderPass(); + }); +} + +VkRenderPass WindowAdaptPass::GetRenderPass() { + return *render_pass; +} + +void WindowAdaptPass::CreateDescriptorPool(size_t num_images) { + const std::array pool_sizes{{ + { + .type = VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER, + .descriptorCount = static_cast(num_images), + }, + { + .type = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + .descriptorCount = static_cast(num_images), + }, + }}; + + const VkDescriptorPoolCreateInfo ci{ + .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .maxSets = static_cast(num_images), + .poolSizeCount = static_cast(pool_sizes.size()), + .pPoolSizes = pool_sizes.data(), + }; + descriptor_pool = device.GetLogical().CreateDescriptorPool(ci); +} + +void WindowAdaptPass::CreateDescriptorSetLayout() { + const std::array layout_bindings{{ + { + .binding = 0, + .descriptorType = VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER, + .descriptorCount = 1, + .stageFlags = VK_SHADER_STAGE_VERTEX_BIT, + .pImmutableSamplers = nullptr, + }, + { + .binding = 1, + .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + .descriptorCount = 1, + .stageFlags = VK_SHADER_STAGE_FRAGMENT_BIT, + .pImmutableSamplers = nullptr, + }, + }}; + + const VkDescriptorSetLayoutCreateInfo ci{ + .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .bindingCount = static_cast(layout_bindings.size()), + .pBindings = layout_bindings.data(), + }; + + descriptor_set_layout = device.GetLogical().CreateDescriptorSetLayout(ci); +} + +void WindowAdaptPass::CreateDescriptorSets(size_t num_images) { + const std::vector layouts(num_images, *descriptor_set_layout); + descriptor_sets = CreateWrappedDescriptorSets(descriptor_pool, layouts); +} + +void WindowAdaptPass::CreateBuffer(const MemoryAllocator& memory_allocator) { + const VkBufferCreateInfo ci{ + .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .size = sizeof(BufferData), + .usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT | + VK_BUFFER_USAGE_VERTEX_BUFFER_BIT | VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT, + .sharingMode = VK_SHARING_MODE_EXCLUSIVE, + .queueFamilyIndexCount = 0, + .pQueueFamilyIndices = nullptr, + }; + + buffer = memory_allocator.CreateBuffer(ci, MemoryUsage::Upload); +} + +void WindowAdaptPass::CreateRenderPass(VkFormat frame_format) { + const VkAttachmentDescription color_attachment{ + .flags = 0, + .format = frame_format, + .samples = VK_SAMPLE_COUNT_1_BIT, + .loadOp = VK_ATTACHMENT_LOAD_OP_CLEAR, + .storeOp = VK_ATTACHMENT_STORE_OP_STORE, + .stencilLoadOp = VK_ATTACHMENT_LOAD_OP_DONT_CARE, + .stencilStoreOp = VK_ATTACHMENT_STORE_OP_DONT_CARE, + .initialLayout = VK_IMAGE_LAYOUT_UNDEFINED, + .finalLayout = VK_IMAGE_LAYOUT_GENERAL, + }; + + const VkAttachmentReference color_attachment_ref{ + .attachment = 0, + .layout = VK_IMAGE_LAYOUT_GENERAL, + }; + + const VkSubpassDescription subpass_description{ + .flags = 0, + .pipelineBindPoint = VK_PIPELINE_BIND_POINT_GRAPHICS, + .inputAttachmentCount = 0, + .pInputAttachments = nullptr, + .colorAttachmentCount = 1, + .pColorAttachments = &color_attachment_ref, + .pResolveAttachments = nullptr, + .pDepthStencilAttachment = nullptr, + .preserveAttachmentCount = 0, + .pPreserveAttachments = nullptr, + }; + + const VkSubpassDependency dependency{ + .srcSubpass = VK_SUBPASS_EXTERNAL, + .dstSubpass = 0, + .srcStageMask = VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT, + .dstStageMask = VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT, + .srcAccessMask = 0, + .dstAccessMask = VK_ACCESS_COLOR_ATTACHMENT_READ_BIT | VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT, + .dependencyFlags = 0, + }; + + const VkRenderPassCreateInfo renderpass_ci{ + .sType = VK_STRUCTURE_TYPE_RENDER_PASS_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .attachmentCount = 1, + .pAttachments = &color_attachment, + .subpassCount = 1, + .pSubpasses = &subpass_description, + .dependencyCount = 1, + .pDependencies = &dependency, + }; + + render_pass = device.GetLogical().CreateRenderPass(renderpass_ci); +} + +void WindowAdaptPass::CreateVertexShader() { + vertex_shader = BuildShader(device, VULKAN_PRESENT_VERT_SPV); +} + +void WindowAdaptPass::CreatePipelineLayout() { + const VkPipelineLayoutCreateInfo ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .setLayoutCount = 1, + .pSetLayouts = descriptor_set_layout.address(), + .pushConstantRangeCount = 0, + .pPushConstantRanges = nullptr, + }; + pipeline_layout = device.GetLogical().CreatePipelineLayout(ci); +} + +void WindowAdaptPass::SetUniformData(BufferData& data, + const Layout::FramebufferLayout& layout) const { + data.uniform.modelview_matrix = + MakeOrthographicMatrix(static_cast(layout.width), static_cast(layout.height)); +} + +void WindowAdaptPass::SetVertexData(BufferData& data, const Layout::FramebufferLayout& layout, + const Common::Rectangle& crop) const { + // Map the coordinates to the screen. + const auto& screen = layout.screen; + const auto x = static_cast(screen.left); + const auto y = static_cast(screen.top); + const auto w = static_cast(screen.GetWidth()); + const auto h = static_cast(screen.GetHeight()); + + data.vertices[0] = ScreenRectVertex(x, y, crop.left, crop.top); + data.vertices[1] = ScreenRectVertex(x + w, y, crop.right, crop.top); + data.vertices[2] = ScreenRectVertex(x, y + h, crop.left, crop.bottom); + data.vertices[3] = ScreenRectVertex(x + w, y + h, crop.right, crop.bottom); +} + +void WindowAdaptPass::UpdateDescriptorSet(size_t image_index, VkImageView image_view) { + const VkDescriptorBufferInfo buffer_info{ + .buffer = *buffer, + .offset = offsetof(BufferData, uniform), + .range = sizeof(BufferData::uniform), + }; + + const VkWriteDescriptorSet ubo_write{ + .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET, + .pNext = nullptr, + .dstSet = descriptor_sets[image_index], + .dstBinding = 0, + .dstArrayElement = 0, + .descriptorCount = 1, + .descriptorType = VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER, + .pImageInfo = nullptr, + .pBufferInfo = &buffer_info, + .pTexelBufferView = nullptr, + }; + + const VkDescriptorImageInfo image_info{ + .sampler = *sampler, + .imageView = image_view, + .imageLayout = VK_IMAGE_LAYOUT_GENERAL, + }; + + const VkWriteDescriptorSet sampler_write{ + .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET, + .pNext = nullptr, + .dstSet = descriptor_sets[image_index], + .dstBinding = 1, + .dstArrayElement = 0, + .descriptorCount = 1, + .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + .pImageInfo = &image_info, + .pBufferInfo = nullptr, + .pTexelBufferView = nullptr, + }; + + device.GetLogical().UpdateDescriptorSets(std::array{ubo_write, sampler_write}, {}); +} + +void WindowAdaptPass::ConfigureLayout(size_t image_index, VkImageView image_view, + const Layout::FramebufferLayout& layout, + const Common::Rectangle& crop_rect) { + BufferData data; + SetUniformData(data, layout); + SetVertexData(data, layout, crop_rect); + + const std::span mapped_span = buffer.Mapped(); + std::memcpy(mapped_span.data(), &data, sizeof(data)); + + UpdateDescriptorSet(image_index, image_view); +} + +void WindowAdaptPass::CreatePipeline() { + const std::array shader_stages{{ + { + .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .stage = VK_SHADER_STAGE_VERTEX_BIT, + .module = *vertex_shader, + .pName = "main", + .pSpecializationInfo = nullptr, + }, + { + .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .stage = VK_SHADER_STAGE_FRAGMENT_BIT, + .module = *fragment_shader, + .pName = "main", + .pSpecializationInfo = nullptr, + }, + }}; + + const auto vertex_binding_description = ScreenRectVertex::GetDescription(); + const auto vertex_attrs_description = ScreenRectVertex::GetAttributes(); + + const VkPipelineVertexInputStateCreateInfo vertex_input_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .vertexBindingDescriptionCount = 1, + .pVertexBindingDescriptions = &vertex_binding_description, + .vertexAttributeDescriptionCount = u32{vertex_attrs_description.size()}, + .pVertexAttributeDescriptions = vertex_attrs_description.data(), + }; + + const VkPipelineInputAssemblyStateCreateInfo input_assembly_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .topology = VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP, + .primitiveRestartEnable = VK_FALSE, + }; + + const VkPipelineViewportStateCreateInfo viewport_state_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .viewportCount = 1, + .pViewports = nullptr, + .scissorCount = 1, + .pScissors = nullptr, + }; + + const VkPipelineRasterizationStateCreateInfo rasterization_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .depthClampEnable = VK_FALSE, + .rasterizerDiscardEnable = VK_FALSE, + .polygonMode = VK_POLYGON_MODE_FILL, + .cullMode = VK_CULL_MODE_NONE, + .frontFace = VK_FRONT_FACE_CLOCKWISE, + .depthBiasEnable = VK_FALSE, + .depthBiasConstantFactor = 0.0f, + .depthBiasClamp = 0.0f, + .depthBiasSlopeFactor = 0.0f, + .lineWidth = 1.0f, + }; + + const VkPipelineMultisampleStateCreateInfo multisampling_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_MULTISAMPLE_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .rasterizationSamples = VK_SAMPLE_COUNT_1_BIT, + .sampleShadingEnable = VK_FALSE, + .minSampleShading = 0.0f, + .pSampleMask = nullptr, + .alphaToCoverageEnable = VK_FALSE, + .alphaToOneEnable = VK_FALSE, + }; + + const VkPipelineColorBlendAttachmentState color_blend_attachment{ + .blendEnable = VK_FALSE, + .srcColorBlendFactor = VK_BLEND_FACTOR_ZERO, + .dstColorBlendFactor = VK_BLEND_FACTOR_ZERO, + .colorBlendOp = VK_BLEND_OP_ADD, + .srcAlphaBlendFactor = VK_BLEND_FACTOR_ZERO, + .dstAlphaBlendFactor = VK_BLEND_FACTOR_ZERO, + .alphaBlendOp = VK_BLEND_OP_ADD, + .colorWriteMask = VK_COLOR_COMPONENT_R_BIT | VK_COLOR_COMPONENT_G_BIT | + VK_COLOR_COMPONENT_B_BIT | VK_COLOR_COMPONENT_A_BIT, + }; + + const VkPipelineColorBlendStateCreateInfo color_blend_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_COLOR_BLEND_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .logicOpEnable = VK_FALSE, + .logicOp = VK_LOGIC_OP_COPY, + .attachmentCount = 1, + .pAttachments = &color_blend_attachment, + .blendConstants = {0.0f, 0.0f, 0.0f, 0.0f}, + }; + + static constexpr std::array dynamic_states{ + VK_DYNAMIC_STATE_VIEWPORT, + VK_DYNAMIC_STATE_SCISSOR, + }; + const VkPipelineDynamicStateCreateInfo dynamic_state_ci{ + .sType = VK_STRUCTURE_TYPE_PIPELINE_DYNAMIC_STATE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .dynamicStateCount = static_cast(dynamic_states.size()), + .pDynamicStates = dynamic_states.data(), + }; + + const VkGraphicsPipelineCreateInfo pipeline_ci{ + .sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .stageCount = static_cast(shader_stages.size()), + .pStages = shader_stages.data(), + .pVertexInputState = &vertex_input_ci, + .pInputAssemblyState = &input_assembly_ci, + .pTessellationState = nullptr, + .pViewportState = &viewport_state_ci, + .pRasterizationState = &rasterization_ci, + .pMultisampleState = &multisampling_ci, + .pDepthStencilState = nullptr, + .pColorBlendState = &color_blend_ci, + .pDynamicState = &dynamic_state_ci, + .layout = *pipeline_layout, + .renderPass = *render_pass, + .subpass = 0, + .basePipelineHandle = 0, + .basePipelineIndex = 0, + }; + + pipeline = device.GetLogical().CreateGraphicsPipeline(pipeline_ci); +} + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/window_adapt_pass.h b/src/video_core/renderer_vulkan/present/window_adapt_pass.h new file mode 100644 index 000000000..5309233a2 --- /dev/null +++ b/src/video_core/renderer_vulkan/present/window_adapt_pass.h @@ -0,0 +1,71 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include "common/math_util.h" +#include "video_core/vulkan_common/vulkan_wrapper.h" + +namespace Layout { +struct FramebufferLayout; +} + +namespace Tegra { +struct FramebufferConfig; +} + +namespace Vulkan { + +class Device; +struct Frame; +class MemoryAllocator; +class Scheduler; + +class WindowAdaptPass final { +public: + explicit WindowAdaptPass(const Device& device, const MemoryAllocator& memory_allocator, + size_t num_images, VkFormat frame_format, vk::Sampler&& sampler, + vk::ShaderModule&& fragment_shader); + ~WindowAdaptPass(); + + void Draw(Scheduler& scheduler, size_t image_index, VkImageView src_image_view, + VkExtent2D src_image_extent, const Common::Rectangle& crop_rect, + const Layout::FramebufferLayout& layout, Frame* dst); + + VkRenderPass GetRenderPass(); + +private: + struct BufferData; + + void SetUniformData(BufferData& data, const Layout::FramebufferLayout& layout) const; + void SetVertexData(BufferData& data, const Layout::FramebufferLayout& layout, + const Common::Rectangle& crop_rect) const; + void UpdateDescriptorSet(size_t image_index, VkImageView image_view); + void ConfigureLayout(size_t image_index, VkImageView image_view, + const Layout::FramebufferLayout& layout, + const Common::Rectangle& crop_rect); + + void CreateDescriptorPool(size_t num_images); + void CreateDescriptorSetLayout(); + void CreateDescriptorSets(size_t num_images); + void CreatePipelineLayout(); + void CreateVertexShader(); + void CreateRenderPass(VkFormat frame_format); + void CreatePipeline(); + void CreateBuffer(const MemoryAllocator& memory_allocator); + +private: + const Device& device; + vk::DescriptorPool descriptor_pool; + vk::DescriptorSetLayout descriptor_set_layout; + vk::DescriptorSets descriptor_sets; + vk::PipelineLayout pipeline_layout; + vk::Sampler sampler; + vk::ShaderModule vertex_shader; + vk::ShaderModule fragment_shader; + vk::RenderPass render_pass; + vk::Pipeline pipeline; + vk::Buffer buffer; +}; + +} // namespace Vulkan -- cgit v1.2.3 From 60ee29aac386c31b9a9add6a12f051273fa45dae Mon Sep 17 00:00:00 2001 From: Liam Date: Mon, 15 Jan 2024 00:09:34 -0500 Subject: renderer_opengl: split out FXAA --- src/video_core/renderer_vulkan/present/fxaa.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'src/video_core/renderer_vulkan/present') diff --git a/src/video_core/renderer_vulkan/present/fxaa.cpp b/src/video_core/renderer_vulkan/present/fxaa.cpp index 6f87ddebb..6c772ada3 100644 --- a/src/video_core/renderer_vulkan/present/fxaa.cpp +++ b/src/video_core/renderer_vulkan/present/fxaa.cpp @@ -133,7 +133,7 @@ VkImageView FXAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_ BeginRenderPass(cmdbuf, renderpass, framebuffer, extent); cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline); cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, layout, 0, descriptor_set, {}); - cmdbuf.Draw(4, 1, 0, 0); + cmdbuf.Draw(3, 1, 0, 0); cmdbuf.EndRenderPass(); TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL); }); -- cgit v1.2.3 From 2ed9586130a7b1de6aefc2aede464c4d3430d484 Mon Sep 17 00:00:00 2001 From: Liam Date: Mon, 15 Jan 2024 10:19:02 -0500 Subject: renderer_vulkan: convert FSR to graphics pipeline --- .../renderer_vulkan/present/anti_alias_pass.h | 10 +- src/video_core/renderer_vulkan/present/fsr.cpp | 510 +++++++-------------- src/video_core/renderer_vulkan/present/fsr.h | 71 +-- src/video_core/renderer_vulkan/present/fxaa.cpp | 16 +- src/video_core/renderer_vulkan/present/fxaa.h | 4 +- src/video_core/renderer_vulkan/present/smaa.cpp | 23 +- src/video_core/renderer_vulkan/present/smaa.h | 4 +- src/video_core/renderer_vulkan/present/util.cpp | 37 +- src/video_core/renderer_vulkan/present/util.h | 10 +- 9 files changed, 262 insertions(+), 423 deletions(-) (limited to 'src/video_core/renderer_vulkan/present') diff --git a/src/video_core/renderer_vulkan/present/anti_alias_pass.h b/src/video_core/renderer_vulkan/present/anti_alias_pass.h index c1ec0b9a0..1f20fbd7f 100644 --- a/src/video_core/renderer_vulkan/present/anti_alias_pass.h +++ b/src/video_core/renderer_vulkan/present/anti_alias_pass.h @@ -12,16 +12,14 @@ class Scheduler; class AntiAliasPass { public: virtual ~AntiAliasPass() = default; - virtual VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, - VkImageView source_image_view) = 0; + virtual void Draw(Scheduler& scheduler, size_t image_index, VkImage* inout_image, + VkImageView* inout_image_view) = 0; }; class NoAA final : public AntiAliasPass { public: - virtual VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, - VkImageView source_image_view) { - return source_image_view; - } + void Draw(Scheduler& scheduler, size_t image_index, VkImage* inout_image, + VkImageView* inout_image_view) override {} }; } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/fsr.cpp b/src/video_core/renderer_vulkan/present/fsr.cpp index 30a16a785..3f708be70 100644 --- a/src/video_core/renderer_vulkan/present/fsr.cpp +++ b/src/video_core/renderer_vulkan/present/fsr.cpp @@ -6,11 +6,13 @@ #include "common/settings.h" #include "video_core/fsr.h" -#include "video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp16_comp_spv.h" -#include "video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp32_comp_spv.h" -#include "video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp16_comp_spv.h" -#include "video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp32_comp_spv.h" +#include "video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp16_frag_spv.h" +#include "video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp32_frag_spv.h" +#include "video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp16_frag_spv.h" +#include "video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp32_frag_spv.h" +#include "video_core/host_shaders/vulkan_fidelityfx_fsr_vert_spv.h" #include "video_core/renderer_vulkan/present/fsr.h" +#include "video_core/renderer_vulkan/present/util.h" #include "video_core/renderer_vulkan/vk_scheduler.h" #include "video_core/renderer_vulkan/vk_shader_util.h" #include "video_core/vulkan_common/vulkan_device.h" @@ -18,403 +20,207 @@ namespace Vulkan { using namespace FSR; -FSR::FSR(const Device& device_, MemoryAllocator& memory_allocator_, size_t image_count_, - VkExtent2D output_size_) - : device{device_}, memory_allocator{memory_allocator_}, image_count{image_count_}, - output_size{output_size_} { +using PushConstants = std::array; + +FSR::FSR(const Device& device, MemoryAllocator& memory_allocator, size_t image_count, + VkExtent2D extent) + : m_device{device}, m_memory_allocator{memory_allocator}, + m_image_count{image_count}, m_extent{extent} { CreateImages(); + CreateRenderPasses(); CreateSampler(); CreateShaders(); CreateDescriptorPool(); CreateDescriptorSetLayout(); CreateDescriptorSets(); - CreatePipelineLayout(); - CreatePipeline(); + CreatePipelineLayouts(); + CreatePipelines(); } -VkImageView FSR::Draw(Scheduler& scheduler, size_t image_index, VkImageView image_view, - VkExtent2D input_image_extent, const Common::Rectangle& crop_rect) { - - UpdateDescriptorSet(image_index, image_view); - - scheduler.Record([this, image_index, input_image_extent, crop_rect](vk::CommandBuffer cmdbuf) { - const VkImageMemoryBarrier base_barrier{ - .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER, - .pNext = nullptr, - .srcAccessMask = 0, - .dstAccessMask = 0, - .oldLayout = VK_IMAGE_LAYOUT_GENERAL, - .newLayout = VK_IMAGE_LAYOUT_GENERAL, - .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, - .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, - .image = {}, - .subresourceRange = - { - .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, - .baseMipLevel = 0, - .levelCount = 1, - .baseArrayLayer = 0, - .layerCount = 1, - }, - }; - - cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, *easu_pipeline); - - const f32 input_image_width = static_cast(input_image_extent.width); - const f32 input_image_height = static_cast(input_image_extent.height); - const f32 output_image_width = static_cast(output_size.width); - const f32 output_image_height = static_cast(output_size.height); - const f32 viewport_width = (crop_rect.right - crop_rect.left) * input_image_width; - const f32 viewport_x = crop_rect.left * input_image_width; - const f32 viewport_height = (crop_rect.bottom - crop_rect.top) * input_image_height; - const f32 viewport_y = crop_rect.top * input_image_height; - - std::array push_constants; - FsrEasuConOffset(push_constants.data() + 0, push_constants.data() + 4, - push_constants.data() + 8, push_constants.data() + 12, - - viewport_width, viewport_height, input_image_width, input_image_height, - output_image_width, output_image_height, viewport_x, viewport_y); - cmdbuf.PushConstants(*pipeline_layout, VK_SHADER_STAGE_COMPUTE_BIT, push_constants); - - { - VkImageMemoryBarrier fsr_write_barrier = base_barrier; - fsr_write_barrier.image = *images[image_index]; - fsr_write_barrier.oldLayout = VK_IMAGE_LAYOUT_UNDEFINED; - - cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, - VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0, fsr_write_barrier); - } - - cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, *pipeline_layout, 0, - descriptor_sets[image_index * 2], {}); - cmdbuf.Dispatch(Common::DivCeil(output_size.width, 16u), - Common::DivCeil(output_size.height, 16u), 1); - - cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, *rcas_pipeline); - - const float sharpening = - static_cast(Settings::values.fsr_sharpening_slider.GetValue()) / 100.0f; - - FsrRcasCon(push_constants.data(), sharpening); - cmdbuf.PushConstants(*pipeline_layout, VK_SHADER_STAGE_COMPUTE_BIT, push_constants); - - { - std::array barriers; - auto& fsr_read_barrier = barriers[0]; - auto& blit_write_barrier = barriers[1]; - - fsr_read_barrier = base_barrier; - fsr_read_barrier.image = *images[image_index]; - fsr_read_barrier.srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT; - fsr_read_barrier.dstAccessMask = VK_ACCESS_SHADER_READ_BIT; - - blit_write_barrier = base_barrier; - blit_write_barrier.image = *images[image_count + image_index]; - blit_write_barrier.oldLayout = VK_IMAGE_LAYOUT_UNDEFINED; - blit_write_barrier.newLayout = VK_IMAGE_LAYOUT_GENERAL; - - cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, - VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0, {}, {}, barriers); - } +void FSR::CreateImages() { + m_dynamic_images.resize(m_image_count); + for (auto& images : m_dynamic_images) { + images.images[Easu] = + CreateWrappedImage(m_memory_allocator, m_extent, VK_FORMAT_R16G16B16A16_SFLOAT); + images.images[Rcas] = + CreateWrappedImage(m_memory_allocator, m_extent, VK_FORMAT_R16G16B16A16_SFLOAT); + images.image_views[Easu] = + CreateWrappedImageView(m_device, images.images[Easu], VK_FORMAT_R16G16B16A16_SFLOAT); + images.image_views[Rcas] = + CreateWrappedImageView(m_device, images.images[Rcas], VK_FORMAT_R16G16B16A16_SFLOAT); + } +} - cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, *pipeline_layout, 0, - descriptor_sets[image_index * 2 + 1], {}); - cmdbuf.Dispatch(Common::DivCeil(output_size.width, 16u), - Common::DivCeil(output_size.height, 16u), 1); +void FSR::CreateRenderPasses() { + m_renderpass = CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16B16A16_SFLOAT); - { - std::array barriers; - auto& blit_read_barrier = barriers[0]; + for (auto& images : m_dynamic_images) { + images.framebuffers[Easu] = + CreateWrappedFramebuffer(m_device, m_renderpass, images.image_views[Easu], m_extent); + images.framebuffers[Rcas] = + CreateWrappedFramebuffer(m_device, m_renderpass, images.image_views[Rcas], m_extent); + } +} - blit_read_barrier = base_barrier; - blit_read_barrier.image = *images[image_count + image_index]; - blit_read_barrier.srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT; - blit_read_barrier.dstAccessMask = VK_ACCESS_SHADER_READ_BIT; +void FSR::CreateSampler() { + m_sampler = CreateBilinearSampler(m_device); +} - cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, - VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT, 0, {}, {}, barriers); - } - }); +void FSR::CreateShaders() { + m_vert_shader = BuildShader(m_device, VULKAN_FIDELITYFX_FSR_VERT_SPV); - return *image_views[image_count + image_index]; + if (m_device.IsFloat16Supported()) { + m_easu_shader = BuildShader(m_device, VULKAN_FIDELITYFX_FSR_EASU_FP16_FRAG_SPV); + m_rcas_shader = BuildShader(m_device, VULKAN_FIDELITYFX_FSR_RCAS_FP16_FRAG_SPV); + } else { + m_easu_shader = BuildShader(m_device, VULKAN_FIDELITYFX_FSR_EASU_FP32_FRAG_SPV); + m_rcas_shader = BuildShader(m_device, VULKAN_FIDELITYFX_FSR_RCAS_FP32_FRAG_SPV); + } } void FSR::CreateDescriptorPool() { - const std::array pool_sizes{{ - { - .type = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, - .descriptorCount = static_cast(image_count * 2), - }, - { - .type = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE, - .descriptorCount = static_cast(image_count * 2), - }, - }}; - - const VkDescriptorPoolCreateInfo ci{ - .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .maxSets = static_cast(image_count * 2), - .poolSizeCount = static_cast(pool_sizes.size()), - .pPoolSizes = pool_sizes.data(), - }; - descriptor_pool = device.GetLogical().CreateDescriptorPool(ci); + // EASU: 1 descriptor + // RCAS: 1 descriptor + // 2 descriptors, 2 descriptor sets per invocation + m_descriptor_pool = CreateWrappedDescriptorPool(m_device, 2 * m_image_count, 2 * m_image_count); } void FSR::CreateDescriptorSetLayout() { - const std::array layout_bindings{{ - { - .binding = 0, - .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, - .descriptorCount = 1, - .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT, - .pImmutableSamplers = sampler.address(), - }, - { - .binding = 1, - .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE, - .descriptorCount = 1, - .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT, - .pImmutableSamplers = sampler.address(), - }, - }}; - - const VkDescriptorSetLayoutCreateInfo ci{ - .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .bindingCount = static_cast(layout_bindings.size()), - .pBindings = layout_bindings.data(), - }; - - descriptor_set_layout = device.GetLogical().CreateDescriptorSetLayout(ci); + m_descriptor_set_layout = + CreateWrappedDescriptorSetLayout(m_device, {VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER}); } void FSR::CreateDescriptorSets() { - const u32 sets = static_cast(image_count * 2); - const std::vector layouts(sets, *descriptor_set_layout); - - const VkDescriptorSetAllocateInfo ai{ - .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_ALLOCATE_INFO, - .pNext = nullptr, - .descriptorPool = *descriptor_pool, - .descriptorSetCount = sets, - .pSetLayouts = layouts.data(), - }; + std::vector layouts(MaxFsrStage, *m_descriptor_set_layout); - descriptor_sets = descriptor_pool.Allocate(ai); -} - -void FSR::CreateImages() { - images.resize(image_count * 2); - image_views.resize(image_count * 2); - - for (size_t i = 0; i < image_count * 2; ++i) { - images[i] = memory_allocator.CreateImage(VkImageCreateInfo{ - .sType = VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .imageType = VK_IMAGE_TYPE_2D, - .format = VK_FORMAT_R16G16B16A16_SFLOAT, - .extent = - { - .width = output_size.width, - .height = output_size.height, - .depth = 1, - }, - .mipLevels = 1, - .arrayLayers = 1, - .samples = VK_SAMPLE_COUNT_1_BIT, - .tiling = VK_IMAGE_TILING_OPTIMAL, - .usage = VK_IMAGE_USAGE_TRANSFER_SRC_BIT | VK_IMAGE_USAGE_STORAGE_BIT | - VK_IMAGE_USAGE_SAMPLED_BIT, - .sharingMode = VK_SHARING_MODE_EXCLUSIVE, - .queueFamilyIndexCount = 0, - .pQueueFamilyIndices = nullptr, - .initialLayout = VK_IMAGE_LAYOUT_UNDEFINED, - }); - image_views[i] = device.GetLogical().CreateImageView(VkImageViewCreateInfo{ - .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .image = *images[i], - .viewType = VK_IMAGE_VIEW_TYPE_2D, - .format = VK_FORMAT_R16G16B16A16_SFLOAT, - .components = - { - .r = VK_COMPONENT_SWIZZLE_IDENTITY, - .g = VK_COMPONENT_SWIZZLE_IDENTITY, - .b = VK_COMPONENT_SWIZZLE_IDENTITY, - .a = VK_COMPONENT_SWIZZLE_IDENTITY, - }, - .subresourceRange = - { - .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, - .baseMipLevel = 0, - .levelCount = 1, - .baseArrayLayer = 0, - .layerCount = 1, - }, - }); + for (auto& images : m_dynamic_images) { + images.descriptor_sets = CreateWrappedDescriptorSets(m_descriptor_pool, layouts); } } -void FSR::CreatePipelineLayout() { - VkPushConstantRange push_const{ - .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT, +void FSR::CreatePipelineLayouts() { + const VkPushConstantRange range{ + .stageFlags = VK_SHADER_STAGE_FRAGMENT_BIT, .offset = 0, - .size = sizeof(std::array), + .size = sizeof(PushConstants), }; VkPipelineLayoutCreateInfo ci{ .sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO, .pNext = nullptr, .flags = 0, .setLayoutCount = 1, - .pSetLayouts = descriptor_set_layout.address(), + .pSetLayouts = m_descriptor_set_layout.address(), .pushConstantRangeCount = 1, - .pPushConstantRanges = &push_const, + .pPushConstantRanges = &range, }; - pipeline_layout = device.GetLogical().CreatePipelineLayout(ci); + m_pipeline_layout = m_device.GetLogical().CreatePipelineLayout(ci); } -void FSR::UpdateDescriptorSet(std::size_t image_index, VkImageView image_view) const { - const auto fsr_image_view = *image_views[image_index]; - const auto blit_image_view = *image_views[image_count + image_index]; - - const VkDescriptorImageInfo image_info{ - .sampler = VK_NULL_HANDLE, - .imageView = image_view, - .imageLayout = VK_IMAGE_LAYOUT_GENERAL, - }; - const VkDescriptorImageInfo fsr_image_info{ - .sampler = VK_NULL_HANDLE, - .imageView = fsr_image_view, - .imageLayout = VK_IMAGE_LAYOUT_GENERAL, - }; - const VkDescriptorImageInfo blit_image_info{ - .sampler = VK_NULL_HANDLE, - .imageView = blit_image_view, - .imageLayout = VK_IMAGE_LAYOUT_GENERAL, - }; - - VkWriteDescriptorSet sampler_write{ - .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET, - .pNext = nullptr, - .dstSet = descriptor_sets[image_index * 2], - .dstBinding = 0, - .dstArrayElement = 0, - .descriptorCount = 1, - .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, - .pImageInfo = &image_info, - .pBufferInfo = nullptr, - .pTexelBufferView = nullptr, - }; - - VkWriteDescriptorSet output_write{ - .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET, - .pNext = nullptr, - .dstSet = descriptor_sets[image_index * 2], - .dstBinding = 1, - .dstArrayElement = 0, - .descriptorCount = 1, - .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE, - .pImageInfo = &fsr_image_info, - .pBufferInfo = nullptr, - .pTexelBufferView = nullptr, - }; - - device.GetLogical().UpdateDescriptorSets(std::array{sampler_write, output_write}, {}); - - sampler_write.dstSet = descriptor_sets[image_index * 2 + 1]; - sampler_write.pImageInfo = &fsr_image_info; - output_write.dstSet = descriptor_sets[image_index * 2 + 1]; - output_write.pImageInfo = &blit_image_info; - - device.GetLogical().UpdateDescriptorSets(std::array{sampler_write, output_write}, {}); +void FSR::CreatePipelines() { + m_easu_pipeline = CreateWrappedPipeline(m_device, m_renderpass, m_pipeline_layout, + std::tie(m_vert_shader, m_easu_shader)); + m_rcas_pipeline = CreateWrappedPipeline(m_device, m_renderpass, m_pipeline_layout, + std::tie(m_vert_shader, m_rcas_shader)); } -void FSR::CreateSampler() { - const VkSamplerCreateInfo ci{ - .sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .magFilter = VK_FILTER_LINEAR, - .minFilter = VK_FILTER_LINEAR, - .mipmapMode = VK_SAMPLER_MIPMAP_MODE_LINEAR, - .addressModeU = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE, - .addressModeV = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE, - .addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE, - .mipLodBias = 0.0f, - .anisotropyEnable = VK_FALSE, - .maxAnisotropy = 0.0f, - .compareEnable = VK_FALSE, - .compareOp = VK_COMPARE_OP_NEVER, - .minLod = 0.0f, - .maxLod = 0.0f, - .borderColor = VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK, - .unnormalizedCoordinates = VK_FALSE, - }; +void FSR::UpdateDescriptorSets(VkImageView image_view, size_t image_index) { + Images& images = m_dynamic_images[image_index]; + std::vector image_infos; + std::vector updates; + image_infos.reserve(2); - sampler = device.GetLogical().CreateSampler(ci); -} + updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, image_view, + images.descriptor_sets[Easu], 0)); + updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, *images.image_views[Easu], + images.descriptor_sets[Rcas], 0)); -void FSR::CreateShaders() { - if (device.IsFloat16Supported()) { - easu_shader = BuildShader(device, VULKAN_FIDELITYFX_FSR_EASU_FP16_COMP_SPV); - rcas_shader = BuildShader(device, VULKAN_FIDELITYFX_FSR_RCAS_FP16_COMP_SPV); - } else { - easu_shader = BuildShader(device, VULKAN_FIDELITYFX_FSR_EASU_FP32_COMP_SPV); - rcas_shader = BuildShader(device, VULKAN_FIDELITYFX_FSR_RCAS_FP32_COMP_SPV); - } + m_device.GetLogical().UpdateDescriptorSets(updates, {}); } -void FSR::CreatePipeline() { - VkPipelineShaderStageCreateInfo shader_stage_easu{ - .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .stage = VK_SHADER_STAGE_COMPUTE_BIT, - .module = *easu_shader, - .pName = "main", - .pSpecializationInfo = nullptr, - }; +void FSR::UploadImages(Scheduler& scheduler) { + if (m_images_ready) { + return; + } - VkPipelineShaderStageCreateInfo shader_stage_rcas{ - .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .stage = VK_SHADER_STAGE_COMPUTE_BIT, - .module = *rcas_shader, - .pName = "main", - .pSpecializationInfo = nullptr, - }; + scheduler.Record([&](vk::CommandBuffer cmdbuf) { + for (auto& image : m_dynamic_images) { + ClearColorImage(cmdbuf, *image.images[Easu]); + ClearColorImage(cmdbuf, *image.images[Rcas]); + } + }); + scheduler.Finish(); - VkComputePipelineCreateInfo pipeline_ci_easu{ - .sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .stage = shader_stage_easu, - .layout = *pipeline_layout, - .basePipelineHandle = VK_NULL_HANDLE, - .basePipelineIndex = 0, - }; + m_images_ready = true; +} - VkComputePipelineCreateInfo pipeline_ci_rcas{ - .sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .stage = shader_stage_rcas, - .layout = *pipeline_layout, - .basePipelineHandle = VK_NULL_HANDLE, - .basePipelineIndex = 0, - }; +VkImageView FSR::Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, + VkImageView source_image_view, VkExtent2D input_image_extent, + const Common::Rectangle& crop_rect) { + Images& images = m_dynamic_images[image_index]; + + VkImage easu_image = *images.images[Easu]; + VkImage rcas_image = *images.images[Rcas]; + VkDescriptorSet easu_descriptor_set = images.descriptor_sets[Easu]; + VkDescriptorSet rcas_descriptor_set = images.descriptor_sets[Rcas]; + VkFramebuffer easu_framebuffer = *images.framebuffers[Easu]; + VkFramebuffer rcas_framebuffer = *images.framebuffers[Rcas]; + VkPipeline easu_pipeline = *m_easu_pipeline; + VkPipeline rcas_pipeline = *m_rcas_pipeline; + VkPipelineLayout pipeline_layout = *m_pipeline_layout; + VkRenderPass renderpass = *m_renderpass; + VkExtent2D extent = m_extent; + + const f32 input_image_width = static_cast(input_image_extent.width); + const f32 input_image_height = static_cast(input_image_extent.height); + const f32 output_image_width = static_cast(extent.width); + const f32 output_image_height = static_cast(extent.height); + const f32 viewport_width = (crop_rect.right - crop_rect.left) * input_image_width; + const f32 viewport_x = crop_rect.left * input_image_width; + const f32 viewport_height = (crop_rect.bottom - crop_rect.top) * input_image_height; + const f32 viewport_y = crop_rect.top * input_image_height; + + PushConstants easu_con{}; + PushConstants rcas_con{}; + FsrEasuConOffset(easu_con.data() + 0, easu_con.data() + 4, easu_con.data() + 8, + easu_con.data() + 12, viewport_width, viewport_height, input_image_width, + input_image_height, output_image_width, output_image_height, viewport_x, + viewport_y); + + const float sharpening = + static_cast(Settings::values.fsr_sharpening_slider.GetValue()) / 100.0f; + FsrRcasCon(rcas_con.data(), sharpening); + + UploadImages(scheduler); + UpdateDescriptorSets(source_image_view, image_index); + + scheduler.RequestOutsideRenderPassOperationContext(); + scheduler.Record([=](vk::CommandBuffer cmdbuf) { + TransitionImageLayout(cmdbuf, source_image, VK_IMAGE_LAYOUT_GENERAL); + TransitionImageLayout(cmdbuf, easu_image, VK_IMAGE_LAYOUT_GENERAL); + BeginRenderPass(cmdbuf, renderpass, easu_framebuffer, extent); + cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, easu_pipeline); + cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline_layout, 0, + easu_descriptor_set, {}); + cmdbuf.PushConstants(pipeline_layout, VK_SHADER_STAGE_FRAGMENT_BIT, easu_con); + cmdbuf.Draw(3, 1, 0, 0); + cmdbuf.EndRenderPass(); + + TransitionImageLayout(cmdbuf, easu_image, VK_IMAGE_LAYOUT_GENERAL); + TransitionImageLayout(cmdbuf, rcas_image, VK_IMAGE_LAYOUT_GENERAL); + BeginRenderPass(cmdbuf, renderpass, rcas_framebuffer, extent); + cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, rcas_pipeline); + cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline_layout, 0, + rcas_descriptor_set, {}); + cmdbuf.PushConstants(pipeline_layout, VK_SHADER_STAGE_FRAGMENT_BIT, rcas_con); + cmdbuf.Draw(3, 1, 0, 0); + cmdbuf.EndRenderPass(); + + TransitionImageLayout(cmdbuf, rcas_image, VK_IMAGE_LAYOUT_GENERAL); + }); - easu_pipeline = device.GetLogical().CreateComputePipeline(pipeline_ci_easu); - rcas_pipeline = device.GetLogical().CreateComputePipeline(pipeline_ci_rcas); + return *images.image_views[Rcas]; } } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/fsr.h b/src/video_core/renderer_vulkan/present/fsr.h index 3505c1416..8602e8146 100644 --- a/src/video_core/renderer_vulkan/present/fsr.h +++ b/src/video_core/renderer_vulkan/present/fsr.h @@ -15,38 +15,55 @@ class Scheduler; class FSR { public: explicit FSR(const Device& device, MemoryAllocator& memory_allocator, size_t image_count, - VkExtent2D output_size); - VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImageView image_view, - VkExtent2D input_image_extent, const Common::Rectangle& crop_rect); + VkExtent2D extent); + VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, + VkImageView source_image_view, VkExtent2D input_image_extent, + const Common::Rectangle& crop_rect); private: - void CreateDescriptorPool(); - void CreateDescriptorSetLayout(); - void CreateDescriptorSets(); void CreateImages(); + void CreateRenderPasses(); void CreateSampler(); void CreateShaders(); - void CreatePipeline(); - void CreatePipelineLayout(); - - void UpdateDescriptorSet(std::size_t image_index, VkImageView image_view) const; - - const Device& device; - MemoryAllocator& memory_allocator; - size_t image_count; - VkExtent2D output_size; - - vk::DescriptorPool descriptor_pool; - vk::DescriptorSetLayout descriptor_set_layout; - vk::DescriptorSets descriptor_sets; - vk::PipelineLayout pipeline_layout; - vk::ShaderModule easu_shader; - vk::ShaderModule rcas_shader; - vk::Pipeline easu_pipeline; - vk::Pipeline rcas_pipeline; - vk::Sampler sampler; - std::vector images; - std::vector image_views; + void CreateDescriptorPool(); + void CreateDescriptorSetLayout(); + void CreateDescriptorSets(); + void CreatePipelineLayouts(); + void CreatePipelines(); + + void UploadImages(Scheduler& scheduler); + void UpdateDescriptorSets(VkImageView image_view, size_t image_index); + + const Device& m_device; + MemoryAllocator& m_memory_allocator; + const size_t m_image_count; + const VkExtent2D m_extent; + + enum FsrStage { + Easu, + Rcas, + MaxFsrStage, + }; + + vk::DescriptorPool m_descriptor_pool; + vk::DescriptorSetLayout m_descriptor_set_layout; + vk::PipelineLayout m_pipeline_layout; + vk::ShaderModule m_vert_shader; + vk::ShaderModule m_easu_shader; + vk::ShaderModule m_rcas_shader; + vk::Pipeline m_easu_pipeline; + vk::Pipeline m_rcas_pipeline; + vk::RenderPass m_renderpass; + vk::Sampler m_sampler; + + struct Images { + vk::DescriptorSets descriptor_sets; + std::array images; + std::array image_views; + std::array framebuffers; + }; + std::vector m_dynamic_images; + bool m_images_ready{}; }; } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/fxaa.cpp b/src/video_core/renderer_vulkan/present/fxaa.cpp index 6c772ada3..bdafd1f4d 100644 --- a/src/video_core/renderer_vulkan/present/fxaa.cpp +++ b/src/video_core/renderer_vulkan/present/fxaa.cpp @@ -63,7 +63,9 @@ void FXAA::CreateDescriptorPool() { } void FXAA::CreateDescriptorSetLayouts() { - m_descriptor_set_layout = CreateWrappedDescriptorSetLayout(m_device, 2); + m_descriptor_set_layout = + CreateWrappedDescriptorSetLayout(m_device, {VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER}); } void FXAA::CreateDescriptorSets() { @@ -112,9 +114,10 @@ void FXAA::UploadImages(Scheduler& scheduler) { m_images_ready = true; } -VkImageView FXAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, - VkImageView source_image_view) { +void FXAA::Draw(Scheduler& scheduler, size_t image_index, VkImage* inout_image, + VkImageView* inout_image_view) { const Image& image{m_dynamic_images[image_index]}; + const VkImage input_image{*inout_image}; const VkImage output_image{*image.image}; const VkDescriptorSet descriptor_set{image.descriptor_sets[0]}; const VkFramebuffer framebuffer{*image.framebuffer}; @@ -124,11 +127,11 @@ VkImageView FXAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_ const VkExtent2D extent{m_extent}; UploadImages(scheduler); - UpdateDescriptorSets(source_image_view, image_index); + UpdateDescriptorSets(*inout_image_view, image_index); scheduler.RequestOutsideRenderPassOperationContext(); scheduler.Record([=](vk::CommandBuffer cmdbuf) { - TransitionImageLayout(cmdbuf, source_image, VK_IMAGE_LAYOUT_GENERAL); + TransitionImageLayout(cmdbuf, input_image, VK_IMAGE_LAYOUT_GENERAL); TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL); BeginRenderPass(cmdbuf, renderpass, framebuffer, extent); cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline); @@ -138,7 +141,8 @@ VkImageView FXAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_ TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL); }); - return *image.image_view; + *inout_image = *image.image; + *inout_image_view = *image.image_view; } } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/fxaa.h b/src/video_core/renderer_vulkan/present/fxaa.h index c083f3ff0..97a2e5c1c 100644 --- a/src/video_core/renderer_vulkan/present/fxaa.h +++ b/src/video_core/renderer_vulkan/present/fxaa.h @@ -19,8 +19,8 @@ public: VkExtent2D extent); ~FXAA() override; - VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, - VkImageView source_image_view) override; + void Draw(Scheduler& scheduler, size_t image_index, VkImage* inout_image, + VkImageView* inout_image_view) override; private: void CreateImages(); diff --git a/src/video_core/renderer_vulkan/present/smaa.cpp b/src/video_core/renderer_vulkan/present/smaa.cpp index 68cd22b08..39645fd1d 100644 --- a/src/video_core/renderer_vulkan/present/smaa.cpp +++ b/src/video_core/renderer_vulkan/present/smaa.cpp @@ -122,10 +122,15 @@ void SMAA::CreateDescriptorPool() { } void SMAA::CreateDescriptorSetLayouts() { - m_descriptor_set_layouts[EdgeDetection] = CreateWrappedDescriptorSetLayout(m_device, 1); + m_descriptor_set_layouts[EdgeDetection] = + CreateWrappedDescriptorSetLayout(m_device, {VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER}); m_descriptor_set_layouts[BlendingWeightCalculation] = - CreateWrappedDescriptorSetLayout(m_device, 3); - m_descriptor_set_layouts[NeighborhoodBlending] = CreateWrappedDescriptorSetLayout(m_device, 2); + CreateWrappedDescriptorSetLayout(m_device, {VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER}); + m_descriptor_set_layouts[NeighborhoodBlending] = + CreateWrappedDescriptorSetLayout(m_device, {VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER}); } void SMAA::CreateDescriptorSets() { @@ -204,10 +209,11 @@ void SMAA::UploadImages(Scheduler& scheduler) { m_images_ready = true; } -VkImageView SMAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, - VkImageView source_image_view) { +void SMAA::Draw(Scheduler& scheduler, size_t image_index, VkImage* inout_image, + VkImageView* inout_image_view) { Images& images = m_dynamic_images[image_index]; + VkImage input_image = *inout_image; VkImage output_image = *images.images[Output]; VkImage edges_image = *images.images[Edges]; VkImage blend_image = *images.images[Blend]; @@ -224,11 +230,11 @@ VkImageView SMAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_ VkFramebuffer neighborhood_blending_framebuffer = *images.framebuffers[NeighborhoodBlending]; UploadImages(scheduler); - UpdateDescriptorSets(source_image_view, image_index); + UpdateDescriptorSets(*inout_image_view, image_index); scheduler.RequestOutsideRenderPassOperationContext(); scheduler.Record([=, this](vk::CommandBuffer cmdbuf) { - TransitionImageLayout(cmdbuf, source_image, VK_IMAGE_LAYOUT_GENERAL); + TransitionImageLayout(cmdbuf, input_image, VK_IMAGE_LAYOUT_GENERAL); TransitionImageLayout(cmdbuf, edges_image, VK_IMAGE_LAYOUT_GENERAL); BeginRenderPass(cmdbuf, *m_renderpasses[EdgeDetection], edge_detection_framebuffer, m_extent); @@ -264,7 +270,8 @@ VkImageView SMAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_ TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL); }); - return *images.image_views[Output]; + *inout_image = *images.images[Output]; + *inout_image_view = *images.image_views[Output]; } } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/smaa.h b/src/video_core/renderer_vulkan/present/smaa.h index 3d6707d48..fdf6def07 100644 --- a/src/video_core/renderer_vulkan/present/smaa.h +++ b/src/video_core/renderer_vulkan/present/smaa.h @@ -20,8 +20,8 @@ public: VkExtent2D extent); ~SMAA() override; - VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image, - VkImageView source_image_view) override; + void Draw(Scheduler& scheduler, size_t image_index, VkImage* inout_image, + VkImageView* inout_image_view) override; private: enum SMAAStage { diff --git a/src/video_core/renderer_vulkan/present/util.cpp b/src/video_core/renderer_vulkan/present/util.cpp index cd6061101..9c08ac613 100644 --- a/src/video_core/renderer_vulkan/present/util.cpp +++ b/src/video_core/renderer_vulkan/present/util.cpp @@ -215,32 +215,37 @@ vk::ShaderModule CreateWrappedShaderModule(const Device& device, std::span(max_descriptors), - }; +vk::DescriptorPool CreateWrappedDescriptorPool(const Device& device, size_t max_descriptors, + size_t max_sets, + std::initializer_list types) { + std::vector pool_sizes(types.size()); + for (u32 i = 0; i < types.size(); i++) { + pool_sizes[i] = VkDescriptorPoolSize{ + .type = std::data(types)[i], + .descriptorCount = static_cast(max_descriptors), + }; + } return device.GetLogical().CreateDescriptorPool(VkDescriptorPoolCreateInfo{ .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO, .pNext = nullptr, .flags = 0, - .maxSets = max_sets, - .poolSizeCount = 1, - .pPoolSizes = &pool_size, + .maxSets = static_cast(max_sets), + .poolSizeCount = static_cast(pool_sizes.size()), + .pPoolSizes = pool_sizes.data(), }); } -vk::DescriptorSetLayout CreateWrappedDescriptorSetLayout(const Device& device, - u32 max_sampler_bindings) { - std::vector bindings(max_sampler_bindings); - for (u32 i = 0; i < max_sampler_bindings; i++) { +vk::DescriptorSetLayout CreateWrappedDescriptorSetLayout( + const Device& device, std::initializer_list types) { + std::vector bindings(types.size()); + for (size_t i = 0; i < types.size(); i++) { bindings[i] = { - .binding = i, - .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + .binding = static_cast(i), + .descriptorType = std::data(types)[i], .descriptorCount = 1, - .stageFlags = VK_SHADER_STAGE_VERTEX_BIT | VK_SHADER_STAGE_FRAGMENT_BIT, + .stageFlags = VK_SHADER_STAGE_VERTEX_BIT | VK_SHADER_STAGE_FRAGMENT_BIT | + VK_SHADER_STAGE_COMPUTE_BIT, .pImmutableSamplers = nullptr, }; } diff --git a/src/video_core/renderer_vulkan/present/util.h b/src/video_core/renderer_vulkan/present/util.h index ea9a26c3d..2f3a538fa 100644 --- a/src/video_core/renderer_vulkan/present/util.h +++ b/src/video_core/renderer_vulkan/present/util.h @@ -25,10 +25,12 @@ vk::Framebuffer CreateWrappedFramebuffer(const Device& device, vk::RenderPass& r vk::ImageView& dest_image, VkExtent2D extent); vk::Sampler CreateWrappedSampler(const Device& device, VkFilter filter = VK_FILTER_LINEAR); vk::ShaderModule CreateWrappedShaderModule(const Device& device, std::span code); -vk::DescriptorPool CreateWrappedDescriptorPool(const Device& device, u32 max_sampler_bindings, - u32 max_sets); -vk::DescriptorSetLayout CreateWrappedDescriptorSetLayout(const Device& device, - u32 max_sampler_bindings); +vk::DescriptorPool CreateWrappedDescriptorPool(const Device& device, size_t max_descriptors, + size_t max_sets, + std::initializer_list types = { + VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER}); +vk::DescriptorSetLayout CreateWrappedDescriptorSetLayout( + const Device& device, std::initializer_list types); vk::DescriptorSets CreateWrappedDescriptorSets(vk::DescriptorPool& pool, vk::Span layouts); vk::PipelineLayout CreateWrappedPipelineLayout(const Device& device, -- cgit v1.2.3 From d4de04584f14f3ea8fde4cd79102b887c084fbc2 Mon Sep 17 00:00:00 2001 From: Liam Date: Mon, 15 Jan 2024 15:08:21 -0500 Subject: renderer_opengl: split up blit screen resources into antialias and window adapt passes --- src/video_core/renderer_vulkan/present/window_adapt_pass.cpp | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'src/video_core/renderer_vulkan/present') diff --git a/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp b/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp index 7fd9ecd22..1d1828a4c 100644 --- a/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp +++ b/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp @@ -92,7 +92,9 @@ void WindowAdaptPass::Draw(Scheduler& scheduler, size_t image_index, VkImageView const VkFramebuffer host_framebuffer{*dst->framebuffer}; const VkRenderPass renderpass{*render_pass}; const VkPipeline graphics_pipeline{*pipeline}; + const VkPipelineLayout graphics_pipeline_layout{*pipeline_layout}; const VkDescriptorSet descriptor_set{descriptor_sets[image_index]}; + const VkBuffer vertex_buffer{*buffer}; const VkExtent2D render_area{ .width = dst->width, .height = dst->height, @@ -134,8 +136,8 @@ void WindowAdaptPass::Draw(Scheduler& scheduler, size_t image_index, VkImageView cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, graphics_pipeline); cmdbuf.SetViewport(0, viewport); cmdbuf.SetScissor(0, scissor); - cmdbuf.BindVertexBuffer(0, *buffer, offsetof(BufferData, vertices)); - cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, *pipeline_layout, 0, + cmdbuf.BindVertexBuffer(0, vertex_buffer, offsetof(BufferData, vertices)); + cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, graphics_pipeline_layout, 0, descriptor_set, {}); cmdbuf.Draw(4, 1, 0, 0); cmdbuf.EndRenderPass(); -- cgit v1.2.3 From 9bdf09bd768f73073e9d1cbc65febfd7f7955db3 Mon Sep 17 00:00:00 2001 From: Liam Date: Thu, 18 Jan 2024 11:44:13 -0500 Subject: renderer_vulkan: implement layer stack composition --- src/video_core/renderer_vulkan/present/filters.cpp | 34 +- src/video_core/renderer_vulkan/present/filters.h | 24 +- src/video_core/renderer_vulkan/present/layer.cpp | 336 ++++++++++++++ src/video_core/renderer_vulkan/present/layer.h | 92 ++++ .../present/present_push_constants.h | 34 ++ src/video_core/renderer_vulkan/present/util.cpp | 31 +- src/video_core/renderer_vulkan/present/util.h | 6 +- .../renderer_vulkan/present/window_adapt_pass.cpp | 503 +++------------------ .../renderer_vulkan/present/window_adapt_pass.h | 29 +- 9 files changed, 576 insertions(+), 513 deletions(-) create mode 100644 src/video_core/renderer_vulkan/present/layer.cpp create mode 100644 src/video_core/renderer_vulkan/present/layer.h create mode 100644 src/video_core/renderer_vulkan/present/present_push_constants.h (limited to 'src/video_core/renderer_vulkan/present') diff --git a/src/video_core/renderer_vulkan/present/filters.cpp b/src/video_core/renderer_vulkan/present/filters.cpp index ee6239cc4..b5e08938e 100644 --- a/src/video_core/renderer_vulkan/present/filters.cpp +++ b/src/video_core/renderer_vulkan/present/filters.cpp @@ -27,43 +27,29 @@ vk::ShaderModule SelectScaleForceShader(const Device& device) { } // Anonymous namespace -std::unique_ptr MakeNearestNeighbor(const Device& device, - const MemoryAllocator& memory_allocator, - size_t image_count, VkFormat frame_format) { - return std::make_unique(device, memory_allocator, image_count, frame_format, +std::unique_ptr MakeNearestNeighbor(const Device& device, VkFormat frame_format) { + return std::make_unique(device, frame_format, CreateNearestNeighborSampler(device), BuildShader(device, VULKAN_PRESENT_FRAG_SPV)); } -std::unique_ptr MakeBilinear(const Device& device, - const MemoryAllocator& memory_allocator, - size_t image_count, VkFormat frame_format) { - return std::make_unique(device, memory_allocator, image_count, frame_format, - CreateBilinearSampler(device), +std::unique_ptr MakeBilinear(const Device& device, VkFormat frame_format) { + return std::make_unique(device, frame_format, CreateBilinearSampler(device), BuildShader(device, VULKAN_PRESENT_FRAG_SPV)); } -std::unique_ptr MakeBicubic(const Device& device, - const MemoryAllocator& memory_allocator, - size_t image_count, VkFormat frame_format) { - return std::make_unique(device, memory_allocator, image_count, frame_format, - CreateBilinearSampler(device), +std::unique_ptr MakeBicubic(const Device& device, VkFormat frame_format) { + return std::make_unique(device, frame_format, CreateBilinearSampler(device), BuildShader(device, PRESENT_BICUBIC_FRAG_SPV)); } -std::unique_ptr MakeGaussian(const Device& device, - const MemoryAllocator& memory_allocator, - size_t image_count, VkFormat frame_format) { - return std::make_unique(device, memory_allocator, image_count, frame_format, - CreateBilinearSampler(device), +std::unique_ptr MakeGaussian(const Device& device, VkFormat frame_format) { + return std::make_unique(device, frame_format, CreateBilinearSampler(device), BuildShader(device, PRESENT_GAUSSIAN_FRAG_SPV)); } -std::unique_ptr MakeScaleForce(const Device& device, - const MemoryAllocator& memory_allocator, - size_t image_count, VkFormat frame_format) { - return std::make_unique(device, memory_allocator, image_count, frame_format, - CreateBilinearSampler(device), +std::unique_ptr MakeScaleForce(const Device& device, VkFormat frame_format) { + return std::make_unique(device, frame_format, CreateBilinearSampler(device), SelectScaleForceShader(device)); } diff --git a/src/video_core/renderer_vulkan/present/filters.h b/src/video_core/renderer_vulkan/present/filters.h index 42d7052da..6c83726dd 100644 --- a/src/video_core/renderer_vulkan/present/filters.h +++ b/src/video_core/renderer_vulkan/present/filters.h @@ -7,24 +7,12 @@ namespace Vulkan { -std::unique_ptr MakeNearestNeighbor(const Device& device, - const MemoryAllocator& memory_allocator, - size_t image_count, VkFormat frame_format); +class MemoryAllocator; -std::unique_ptr MakeBilinear(const Device& device, - const MemoryAllocator& memory_allocator, - size_t image_count, VkFormat frame_format); - -std::unique_ptr MakeBicubic(const Device& device, - const MemoryAllocator& memory_allocator, - size_t image_count, VkFormat frame_format); - -std::unique_ptr MakeGaussian(const Device& device, - const MemoryAllocator& memory_allocator, - size_t image_count, VkFormat frame_format); - -std::unique_ptr MakeScaleForce(const Device& device, - const MemoryAllocator& memory_allocator, - size_t image_count, VkFormat frame_format); +std::unique_ptr MakeNearestNeighbor(const Device& device, VkFormat frame_format); +std::unique_ptr MakeBilinear(const Device& device, VkFormat frame_format); +std::unique_ptr MakeBicubic(const Device& device, VkFormat frame_format); +std::unique_ptr MakeGaussian(const Device& device, VkFormat frame_format); +std::unique_ptr MakeScaleForce(const Device& device, VkFormat frame_format); } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/layer.cpp b/src/video_core/renderer_vulkan/present/layer.cpp new file mode 100644 index 000000000..cfc04be44 --- /dev/null +++ b/src/video_core/renderer_vulkan/present/layer.cpp @@ -0,0 +1,336 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#include "video_core/renderer_vulkan/vk_rasterizer.h" + +#include "common/settings.h" +#include "video_core/framebuffer_config.h" +#include "video_core/renderer_vulkan/present/fsr.h" +#include "video_core/renderer_vulkan/present/fxaa.h" +#include "video_core/renderer_vulkan/present/layer.h" +#include "video_core/renderer_vulkan/present/present_push_constants.h" +#include "video_core/renderer_vulkan/present/smaa.h" +#include "video_core/renderer_vulkan/present/util.h" +#include "video_core/renderer_vulkan/vk_blit_screen.h" +#include "video_core/textures/decoders.h" + +namespace Vulkan { + +namespace { + +u32 GetBytesPerPixel(const Tegra::FramebufferConfig& framebuffer) { + using namespace VideoCore::Surface; + return BytesPerBlock(PixelFormatFromGPUPixelFormat(framebuffer.pixel_format)); +} + +std::size_t GetSizeInBytes(const Tegra::FramebufferConfig& framebuffer) { + return static_cast(framebuffer.stride) * + static_cast(framebuffer.height) * GetBytesPerPixel(framebuffer); +} + +VkFormat GetFormat(const Tegra::FramebufferConfig& framebuffer) { + switch (framebuffer.pixel_format) { + case Service::android::PixelFormat::Rgba8888: + case Service::android::PixelFormat::Rgbx8888: + return VK_FORMAT_A8B8G8R8_UNORM_PACK32; + case Service::android::PixelFormat::Rgb565: + return VK_FORMAT_R5G6B5_UNORM_PACK16; + case Service::android::PixelFormat::Bgra8888: + return VK_FORMAT_B8G8R8A8_UNORM; + default: + UNIMPLEMENTED_MSG("Unknown framebuffer pixel format: {}", + static_cast(framebuffer.pixel_format)); + return VK_FORMAT_A8B8G8R8_UNORM_PACK32; + } +} + +} // Anonymous namespace + +Layer::Layer(const Device& device_, MemoryAllocator& memory_allocator_, Scheduler& scheduler_, + Tegra::MaxwellDeviceMemoryManager& device_memory_, size_t image_count_, + VkExtent2D output_size, VkDescriptorSetLayout layout) + : device(device_), memory_allocator(memory_allocator_), scheduler(scheduler_), + device_memory(device_memory_), image_count(image_count_) { + CreateDescriptorPool(); + CreateDescriptorSets(layout); + if (Settings::values.scaling_filter.GetValue() == Settings::ScalingFilter::Fsr) { + CreateFSR(output_size); + } +} + +Layer::~Layer() { + ReleaseRawImages(); +} + +void Layer::ConfigureDraw(PresentPushConstants* out_push_constants, + VkDescriptorSet* out_descriptor_set, RasterizerVulkan& rasterizer, + VkSampler sampler, size_t image_index, + const Tegra::FramebufferConfig& framebuffer, + const Layout::FramebufferLayout& layout) { + const auto texture_info = rasterizer.AccelerateDisplay( + framebuffer, framebuffer.address + framebuffer.offset, framebuffer.stride); + const u32 texture_width = texture_info ? texture_info->width : framebuffer.width; + const u32 texture_height = texture_info ? texture_info->height : framebuffer.height; + const u32 scaled_width = texture_info ? texture_info->scaled_width : texture_width; + const u32 scaled_height = texture_info ? texture_info->scaled_height : texture_height; + const bool use_accelerated = texture_info.has_value(); + + RefreshResources(framebuffer); + SetAntiAliasPass(); + + // Finish any pending renderpass + scheduler.RequestOutsideRenderPassOperationContext(); + scheduler.Wait(resource_ticks[image_index]); + SCOPE_EXIT({ resource_ticks[image_index] = scheduler.CurrentTick(); }); + + if (!use_accelerated) { + UpdateRawImage(framebuffer, image_index); + } + + VkImage source_image = texture_info ? texture_info->image : *raw_images[image_index]; + VkImageView source_image_view = + texture_info ? texture_info->image_view : *raw_image_views[image_index]; + + anti_alias->Draw(scheduler, image_index, &source_image, &source_image_view); + + auto crop_rect = Tegra::NormalizeCrop(framebuffer, texture_width, texture_height); + const VkExtent2D render_extent{ + .width = scaled_width, + .height = scaled_height, + }; + + if (fsr) { + source_image_view = fsr->Draw(scheduler, image_index, source_image, source_image_view, + render_extent, crop_rect); + crop_rect = {0, 0, 1, 1}; + } + + SetMatrixData(*out_push_constants, layout); + SetVertexData(*out_push_constants, layout, crop_rect); + + UpdateDescriptorSet(source_image_view, sampler, image_index); + *out_descriptor_set = descriptor_sets[image_index]; +} + +void Layer::CreateDescriptorPool() { + descriptor_pool = CreateWrappedDescriptorPool(device, image_count, image_count); +} + +void Layer::CreateDescriptorSets(VkDescriptorSetLayout layout) { + const std::vector layouts(image_count, layout); + descriptor_sets = CreateWrappedDescriptorSets(descriptor_pool, layouts); +} + +void Layer::CreateStagingBuffer(const Tegra::FramebufferConfig& framebuffer) { + const VkBufferCreateInfo ci{ + .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .size = CalculateBufferSize(framebuffer), + .usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT | + VK_BUFFER_USAGE_VERTEX_BUFFER_BIT | VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT, + .sharingMode = VK_SHARING_MODE_EXCLUSIVE, + .queueFamilyIndexCount = 0, + .pQueueFamilyIndices = nullptr, + }; + + buffer = memory_allocator.CreateBuffer(ci, MemoryUsage::Upload); +} + +void Layer::CreateRawImages(const Tegra::FramebufferConfig& framebuffer) { + const auto format = GetFormat(framebuffer); + resource_ticks.resize(image_count); + raw_images.resize(image_count); + raw_image_views.resize(image_count); + + for (size_t i = 0; i < image_count; ++i) { + raw_images[i] = + CreateWrappedImage(memory_allocator, {framebuffer.width, framebuffer.height}, format); + raw_image_views[i] = CreateWrappedImageView(device, raw_images[i], format); + } +} + +void Layer::CreateFSR(VkExtent2D output_size) { + fsr = std::make_unique(device, memory_allocator, image_count, output_size); +} + +void Layer::RefreshResources(const Tegra::FramebufferConfig& framebuffer) { + if (framebuffer.width == raw_width && framebuffer.height == raw_height && + framebuffer.pixel_format == pixel_format && !raw_images.empty()) { + return; + } + + raw_width = framebuffer.width; + raw_height = framebuffer.height; + pixel_format = framebuffer.pixel_format; + anti_alias.reset(); + + ReleaseRawImages(); + CreateStagingBuffer(framebuffer); + CreateRawImages(framebuffer); +} + +void Layer::SetAntiAliasPass() { + if (anti_alias && anti_alias_setting == Settings::values.anti_aliasing.GetValue()) { + return; + } + + anti_alias_setting = Settings::values.anti_aliasing.GetValue(); + + const VkExtent2D render_area{ + .width = Settings::values.resolution_info.ScaleUp(raw_width), + .height = Settings::values.resolution_info.ScaleUp(raw_height), + }; + + switch (anti_alias_setting) { + case Settings::AntiAliasing::Fxaa: + anti_alias = std::make_unique(device, memory_allocator, image_count, render_area); + break; + case Settings::AntiAliasing::Smaa: + anti_alias = std::make_unique(device, memory_allocator, image_count, render_area); + break; + default: + anti_alias = std::make_unique(); + break; + } +} + +void Layer::ReleaseRawImages() { + for (const u64 tick : resource_ticks) { + scheduler.Wait(tick); + } + raw_images.clear(); + buffer.reset(); +} + +u64 Layer::CalculateBufferSize(const Tegra::FramebufferConfig& framebuffer) const { + return GetSizeInBytes(framebuffer) * image_count; +} + +u64 Layer::GetRawImageOffset(const Tegra::FramebufferConfig& framebuffer, + size_t image_index) const { + return GetSizeInBytes(framebuffer) * image_index; +} + +void Layer::SetMatrixData(PresentPushConstants& data, + const Layout::FramebufferLayout& layout) const { + data.modelview_matrix = + MakeOrthographicMatrix(static_cast(layout.width), static_cast(layout.height)); +} + +void Layer::SetVertexData(PresentPushConstants& data, const Layout::FramebufferLayout& layout, + const Common::Rectangle& crop) const { + // Map the coordinates to the screen. + const auto& screen = layout.screen; + const auto x = static_cast(screen.left); + const auto y = static_cast(screen.top); + const auto w = static_cast(screen.GetWidth()); + const auto h = static_cast(screen.GetHeight()); + + data.vertices[0] = ScreenRectVertex(x, y, crop.left, crop.top); + data.vertices[1] = ScreenRectVertex(x + w, y, crop.right, crop.top); + data.vertices[2] = ScreenRectVertex(x, y + h, crop.left, crop.bottom); + data.vertices[3] = ScreenRectVertex(x + w, y + h, crop.right, crop.bottom); +} + +void Layer::UpdateDescriptorSet(VkImageView image_view, VkSampler sampler, size_t image_index) { + const VkDescriptorImageInfo image_info{ + .sampler = sampler, + .imageView = image_view, + .imageLayout = VK_IMAGE_LAYOUT_GENERAL, + }; + + const VkWriteDescriptorSet sampler_write{ + .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET, + .pNext = nullptr, + .dstSet = descriptor_sets[image_index], + .dstBinding = 0, + .dstArrayElement = 0, + .descriptorCount = 1, + .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, + .pImageInfo = &image_info, + .pBufferInfo = nullptr, + .pTexelBufferView = nullptr, + }; + + device.GetLogical().UpdateDescriptorSets(std::array{sampler_write}, {}); +} + +void Layer::UpdateRawImage(const Tegra::FramebufferConfig& framebuffer, size_t image_index) { + const std::span mapped_span = buffer.Mapped(); + + const u64 image_offset = GetRawImageOffset(framebuffer, image_index); + + const DAddr framebuffer_addr = framebuffer.address + framebuffer.offset; + const u8* const host_ptr = device_memory.GetPointer(framebuffer_addr); + + // TODO(Rodrigo): Read this from HLE + constexpr u32 block_height_log2 = 4; + const u32 bytes_per_pixel = GetBytesPerPixel(framebuffer); + const u64 linear_size{GetSizeInBytes(framebuffer)}; + const u64 tiled_size{Tegra::Texture::CalculateSize( + true, bytes_per_pixel, framebuffer.stride, framebuffer.height, 1, block_height_log2, 0)}; + Tegra::Texture::UnswizzleTexture( + mapped_span.subspan(image_offset, linear_size), std::span(host_ptr, tiled_size), + bytes_per_pixel, framebuffer.width, framebuffer.height, 1, block_height_log2, 0); + + const VkBufferImageCopy copy{ + .bufferOffset = image_offset, + .bufferRowLength = 0, + .bufferImageHeight = 0, + .imageSubresource = + { + .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .mipLevel = 0, + .baseArrayLayer = 0, + .layerCount = 1, + }, + .imageOffset = {.x = 0, .y = 0, .z = 0}, + .imageExtent = + { + .width = framebuffer.width, + .height = framebuffer.height, + .depth = 1, + }, + }; + scheduler.Record([this, copy, index = image_index](vk::CommandBuffer cmdbuf) { + const VkImage image = *raw_images[index]; + const VkImageMemoryBarrier base_barrier{ + .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER, + .pNext = nullptr, + .srcAccessMask = 0, + .dstAccessMask = 0, + .oldLayout = VK_IMAGE_LAYOUT_GENERAL, + .newLayout = VK_IMAGE_LAYOUT_GENERAL, + .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, + .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, + .image = image, + .subresourceRange{ + .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .baseMipLevel = 0, + .levelCount = 1, + .baseArrayLayer = 0, + .layerCount = 1, + }, + }; + VkImageMemoryBarrier read_barrier = base_barrier; + read_barrier.dstAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT; + read_barrier.oldLayout = VK_IMAGE_LAYOUT_UNDEFINED; + read_barrier.newLayout = VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL; + + VkImageMemoryBarrier write_barrier = base_barrier; + write_barrier.srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT; + write_barrier.dstAccessMask = VK_ACCESS_SHADER_READ_BIT; + write_barrier.oldLayout = VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL; + + cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_HOST_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT, 0, + read_barrier); + cmdbuf.CopyBufferToImage(*buffer, image, VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, copy); + cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, + VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT | + VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, + 0, write_barrier); + }); +} + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/layer.h b/src/video_core/renderer_vulkan/present/layer.h new file mode 100644 index 000000000..88d43fc5f --- /dev/null +++ b/src/video_core/renderer_vulkan/present/layer.h @@ -0,0 +1,92 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include "common/math_util.h" +#include "video_core/host1x/gpu_device_memory_manager.h" +#include "video_core/vulkan_common/vulkan_wrapper.h" + +namespace Layout { +struct FramebufferLayout; +} + +namespace Tegra { +struct FramebufferConfig; +} + +namespace Service::android { +enum class PixelFormat : u32; +} + +namespace Settings { +enum class AntiAliasing : u32; +} + +namespace Vulkan { + +class AntiAliasPass; +class Device; +class FSR; +class MemoryAllocator; +struct PresentPushConstants; +class RasterizerVulkan; +class Scheduler; + +class Layer final { +public: + explicit Layer(const Device& device, MemoryAllocator& memory_allocator, Scheduler& scheduler, + Tegra::MaxwellDeviceMemoryManager& device_memory, size_t image_count, + VkExtent2D output_size, VkDescriptorSetLayout layout); + ~Layer(); + + void ConfigureDraw(PresentPushConstants* out_push_constants, + VkDescriptorSet* out_descriptor_set, RasterizerVulkan& rasterizer, + VkSampler sampler, size_t image_index, + const Tegra::FramebufferConfig& framebuffer, + const Layout::FramebufferLayout& layout); + +private: + void CreateDescriptorPool(); + void CreateDescriptorSets(VkDescriptorSetLayout layout); + void CreateStagingBuffer(const Tegra::FramebufferConfig& framebuffer); + void CreateRawImages(const Tegra::FramebufferConfig& framebuffer); + void CreateFSR(VkExtent2D output_size); + + void RefreshResources(const Tegra::FramebufferConfig& framebuffer); + void SetAntiAliasPass(); + void ReleaseRawImages(); + + u64 CalculateBufferSize(const Tegra::FramebufferConfig& framebuffer) const; + u64 GetRawImageOffset(const Tegra::FramebufferConfig& framebuffer, size_t image_index) const; + + void SetMatrixData(PresentPushConstants& data, const Layout::FramebufferLayout& layout) const; + void SetVertexData(PresentPushConstants& data, const Layout::FramebufferLayout& layout, + const Common::Rectangle& crop) const; + void UpdateDescriptorSet(VkImageView image_view, VkSampler sampler, size_t image_index); + void UpdateRawImage(const Tegra::FramebufferConfig& framebuffer, size_t image_index); + +private: + const Device& device; + MemoryAllocator& memory_allocator; + Scheduler& scheduler; + Tegra::MaxwellDeviceMemoryManager& device_memory; + const size_t image_count{}; + vk::DescriptorPool descriptor_pool{}; + vk::DescriptorSets descriptor_sets{}; + + vk::Buffer buffer{}; + std::vector raw_images{}; + std::vector raw_image_views{}; + u32 raw_width{}; + u32 raw_height{}; + Service::android::PixelFormat pixel_format{}; + + Settings::AntiAliasing anti_alias_setting{}; + std::unique_ptr anti_alias{}; + + std::unique_ptr fsr{}; + std::vector resource_ticks{}; +}; + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/present_push_constants.h b/src/video_core/renderer_vulkan/present/present_push_constants.h new file mode 100644 index 000000000..f1949e7aa --- /dev/null +++ b/src/video_core/renderer_vulkan/present/present_push_constants.h @@ -0,0 +1,34 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include "common/common_types.h" + +namespace Vulkan { + +struct ScreenRectVertex { + ScreenRectVertex() = default; + explicit ScreenRectVertex(f32 x, f32 y, f32 u, f32 v) : position{{x, y}}, tex_coord{{u, v}} {} + + std::array position; + std::array tex_coord; +}; + +static inline std::array MakeOrthographicMatrix(f32 width, f32 height) { + // clang-format off + return { 2.f / width, 0.f, 0.f, 0.f, + 0.f, 2.f / height, 0.f, 0.f, + 0.f, 0.f, 1.f, 0.f, + -1.f, -1.f, 0.f, 1.f}; + // clang-format on +} + +struct PresentPushConstants { + std::array modelview_matrix; + std::array vertices; +}; + +static_assert(sizeof(PresentPushConstants) <= 128, "Push constants are too large"); + +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/util.cpp b/src/video_core/renderer_vulkan/present/util.cpp index 9c08ac613..7bff1c436 100644 --- a/src/video_core/renderer_vulkan/present/util.cpp +++ b/src/video_core/renderer_vulkan/present/util.cpp @@ -113,16 +113,18 @@ vk::ImageView CreateWrappedImageView(const Device& device, vk::Image& image, VkF }); } -vk::RenderPass CreateWrappedRenderPass(const Device& device, VkFormat format) { +vk::RenderPass CreateWrappedRenderPass(const Device& device, VkFormat format, + VkImageLayout initial_layout) { const VkAttachmentDescription attachment{ .flags = VK_ATTACHMENT_DESCRIPTION_MAY_ALIAS_BIT, .format = format, .samples = VK_SAMPLE_COUNT_1_BIT, - .loadOp = VK_ATTACHMENT_LOAD_OP_LOAD, + .loadOp = initial_layout == VK_IMAGE_LAYOUT_UNDEFINED ? VK_ATTACHMENT_LOAD_OP_DONT_CARE + : VK_ATTACHMENT_LOAD_OP_LOAD, .storeOp = VK_ATTACHMENT_STORE_OP_STORE, .stencilLoadOp = VK_ATTACHMENT_LOAD_OP_LOAD, .stencilStoreOp = VK_ATTACHMENT_STORE_OP_STORE, - .initialLayout = VK_IMAGE_LAYOUT_GENERAL, + .initialLayout = initial_layout, .finalLayout = VK_IMAGE_LAYOUT_GENERAL, }; @@ -244,8 +246,7 @@ vk::DescriptorSetLayout CreateWrappedDescriptorSetLayout( .binding = static_cast(i), .descriptorType = std::data(types)[i], .descriptorCount = 1, - .stageFlags = VK_SHADER_STAGE_VERTEX_BIT | VK_SHADER_STAGE_FRAGMENT_BIT | - VK_SHADER_STAGE_COMPUTE_BIT, + .stageFlags = VK_SHADER_STAGE_VERTEX_BIT | VK_SHADER_STAGE_FRAGMENT_BIT, .pImmutableSamplers = nullptr, }; } @@ -285,7 +286,8 @@ vk::PipelineLayout CreateWrappedPipelineLayout(const Device& device, vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderpass, vk::PipelineLayout& layout, - std::tuple shaders) { + std::tuple shaders, + bool enable_blending) { const std::array shader_stages{{ { .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, @@ -363,7 +365,7 @@ vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderp .alphaToOneEnable = VK_FALSE, }; - constexpr VkPipelineColorBlendAttachmentState color_blend_attachment{ + constexpr VkPipelineColorBlendAttachmentState color_blend_attachment_disabled{ .blendEnable = VK_FALSE, .srcColorBlendFactor = VK_BLEND_FACTOR_ZERO, .dstColorBlendFactor = VK_BLEND_FACTOR_ZERO, @@ -375,6 +377,18 @@ vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderp VK_COLOR_COMPONENT_B_BIT | VK_COLOR_COMPONENT_A_BIT, }; + constexpr VkPipelineColorBlendAttachmentState color_blend_attachment_enabled{ + .blendEnable = VK_TRUE, + .srcColorBlendFactor = VK_BLEND_FACTOR_SRC_ALPHA, + .dstColorBlendFactor = VK_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA, + .colorBlendOp = VK_BLEND_OP_ADD, + .srcAlphaBlendFactor = VK_BLEND_FACTOR_ONE, + .dstAlphaBlendFactor = VK_BLEND_FACTOR_ZERO, + .alphaBlendOp = VK_BLEND_OP_ADD, + .colorWriteMask = VK_COLOR_COMPONENT_R_BIT | VK_COLOR_COMPONENT_G_BIT | + VK_COLOR_COMPONENT_B_BIT | VK_COLOR_COMPONENT_A_BIT, + }; + const VkPipelineColorBlendStateCreateInfo color_blend_ci{ .sType = VK_STRUCTURE_TYPE_PIPELINE_COLOR_BLEND_STATE_CREATE_INFO, .pNext = nullptr, @@ -382,7 +396,8 @@ vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderp .logicOpEnable = VK_FALSE, .logicOp = VK_LOGIC_OP_COPY, .attachmentCount = 1, - .pAttachments = &color_blend_attachment, + .pAttachments = + enable_blending ? &color_blend_attachment_enabled : &color_blend_attachment_disabled, .blendConstants = {0.0f, 0.0f, 0.0f, 0.0f}, }; diff --git a/src/video_core/renderer_vulkan/present/util.h b/src/video_core/renderer_vulkan/present/util.h index 2f3a538fa..fb4e4a8e4 100644 --- a/src/video_core/renderer_vulkan/present/util.h +++ b/src/video_core/renderer_vulkan/present/util.h @@ -20,7 +20,8 @@ void UploadImage(const Device& device, MemoryAllocator& allocator, Scheduler& sc void ClearColorImage(vk::CommandBuffer& cmdbuf, VkImage image); vk::ImageView CreateWrappedImageView(const Device& device, vk::Image& image, VkFormat format); -vk::RenderPass CreateWrappedRenderPass(const Device& device, VkFormat format); +vk::RenderPass CreateWrappedRenderPass(const Device& device, VkFormat format, + VkImageLayout initial_layout = VK_IMAGE_LAYOUT_GENERAL); vk::Framebuffer CreateWrappedFramebuffer(const Device& device, vk::RenderPass& render_pass, vk::ImageView& dest_image, VkExtent2D extent); vk::Sampler CreateWrappedSampler(const Device& device, VkFilter filter = VK_FILTER_LINEAR); @@ -37,7 +38,8 @@ vk::PipelineLayout CreateWrappedPipelineLayout(const Device& device, vk::DescriptorSetLayout& layout); vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderpass, vk::PipelineLayout& layout, - std::tuple shaders); + std::tuple shaders, + bool enable_blending = false); VkWriteDescriptorSet CreateWriteDescriptorSet(std::vector& images, VkSampler sampler, VkImageView view, VkDescriptorSet set, u32 binding); diff --git a/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp b/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp index 1d1828a4c..c5db0230d 100644 --- a/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp +++ b/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp @@ -1,10 +1,11 @@ // SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project // SPDX-License-Identifier: GPL-2.0-or-later -#include - #include "core/frontend/framebuffer_layout.h" +#include "video_core/framebuffer_config.h" #include "video_core/host_shaders/vulkan_present_vert_spv.h" +#include "video_core/renderer_vulkan/present/layer.h" +#include "video_core/renderer_vulkan/present/present_push_constants.h" #include "video_core/renderer_vulkan/present/util.h" #include "video_core/renderer_vulkan/present/window_adapt_pass.h" #include "video_core/renderer_vulkan/vk_present_manager.h" @@ -14,501 +15,123 @@ namespace Vulkan { -namespace { - -struct ScreenRectVertex { - ScreenRectVertex() = default; - explicit ScreenRectVertex(f32 x, f32 y, f32 u, f32 v) : position{{x, y}}, tex_coord{{u, v}} {} - - std::array position; - std::array tex_coord; - - static VkVertexInputBindingDescription GetDescription() { - return { - .binding = 0, - .stride = sizeof(ScreenRectVertex), - .inputRate = VK_VERTEX_INPUT_RATE_VERTEX, - }; - } - - static std::array GetAttributes() { - return {{ - { - .location = 0, - .binding = 0, - .format = VK_FORMAT_R32G32_SFLOAT, - .offset = offsetof(ScreenRectVertex, position), - }, - { - .location = 1, - .binding = 0, - .format = VK_FORMAT_R32G32_SFLOAT, - .offset = offsetof(ScreenRectVertex, tex_coord), - }, - }}; - } -}; - -std::array MakeOrthographicMatrix(f32 width, f32 height) { - // clang-format off - return { 2.f / width, 0.f, 0.f, 0.f, - 0.f, 2.f / height, 0.f, 0.f, - 0.f, 0.f, 1.f, 0.f, - -1.f, -1.f, 0.f, 1.f}; - // clang-format on -} - -} // Anonymous namespace - -struct WindowAdaptPass::BufferData { - struct { - std::array modelview_matrix; - } uniform; - - std::array vertices; -}; - -WindowAdaptPass::WindowAdaptPass(const Device& device_, const MemoryAllocator& memory_allocator, - size_t num_images, VkFormat frame_format, vk::Sampler&& sampler_, - vk::ShaderModule&& fragment_shader_) +WindowAdaptPass::WindowAdaptPass(const Device& device_, VkFormat frame_format, + vk::Sampler&& sampler_, vk::ShaderModule&& fragment_shader_) : device(device_), sampler(std::move(sampler_)), fragment_shader(std::move(fragment_shader_)) { - CreateDescriptorPool(num_images); CreateDescriptorSetLayout(); - CreateDescriptorSets(num_images); CreatePipelineLayout(); CreateVertexShader(); CreateRenderPass(frame_format); CreatePipeline(); - CreateBuffer(memory_allocator); } WindowAdaptPass::~WindowAdaptPass() = default; -void WindowAdaptPass::Draw(Scheduler& scheduler, size_t image_index, VkImageView src_image_view, - VkExtent2D src_image_extent, const Common::Rectangle& crop_rect, +void WindowAdaptPass::Draw(RasterizerVulkan& rasterizer, Scheduler& scheduler, size_t image_index, + std::list& layers, + std::span configs, const Layout::FramebufferLayout& layout, Frame* dst) { - ConfigureLayout(image_index, src_image_view, layout, crop_rect); const VkFramebuffer host_framebuffer{*dst->framebuffer}; const VkRenderPass renderpass{*render_pass}; const VkPipeline graphics_pipeline{*pipeline}; const VkPipelineLayout graphics_pipeline_layout{*pipeline_layout}; - const VkDescriptorSet descriptor_set{descriptor_sets[image_index]}; - const VkBuffer vertex_buffer{*buffer}; const VkExtent2D render_area{ .width = dst->width, .height = dst->height, }; + const size_t layer_count = configs.size(); + std::vector push_constants(layer_count); + std::vector descriptor_sets(layer_count); + + auto layer_it = layers.begin(); + for (size_t i = 0; i < layer_count; i++) { + layer_it->ConfigureDraw(&push_constants[i], &descriptor_sets[i], rasterizer, *sampler, + image_index, configs[i], layout); + layer_it++; + } + scheduler.Record([=](vk::CommandBuffer cmdbuf) { const f32 bg_red = Settings::values.bg_red.GetValue() / 255.0f; const f32 bg_green = Settings::values.bg_green.GetValue() / 255.0f; const f32 bg_blue = Settings::values.bg_blue.GetValue() / 255.0f; - const VkClearValue clear_color{ - .color = {.float32 = {bg_red, bg_green, bg_blue, 1.0f}}, + const VkClearAttachment clear_attachment{ + .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .colorAttachment = 0, + .clearValue = + { + .color = {.float32 = {bg_red, bg_green, bg_blue, 1.0f}}, + }, }; - const VkRenderPassBeginInfo renderpass_bi{ - .sType = VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO, - .pNext = nullptr, - .renderPass = renderpass, - .framebuffer = host_framebuffer, - .renderArea = + const VkClearRect clear_rect{ + .rect = { .offset = {0, 0}, .extent = render_area, }, - .clearValueCount = 1, - .pClearValues = &clear_color, - }; - const VkViewport viewport{ - .x = 0.0f, - .y = 0.0f, - .width = static_cast(render_area.width), - .height = static_cast(render_area.height), - .minDepth = 0.0f, - .maxDepth = 1.0f, + .baseArrayLayer = 0, + .layerCount = 1, }; - const VkRect2D scissor{ - .offset = {0, 0}, - .extent = render_area, - }; - cmdbuf.BeginRenderPass(renderpass_bi, VK_SUBPASS_CONTENTS_INLINE); + + BeginRenderPass(cmdbuf, renderpass, host_framebuffer, render_area); + cmdbuf.ClearAttachments({clear_attachment}, {clear_rect}); + cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, graphics_pipeline); - cmdbuf.SetViewport(0, viewport); - cmdbuf.SetScissor(0, scissor); - cmdbuf.BindVertexBuffer(0, vertex_buffer, offsetof(BufferData, vertices)); - cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, graphics_pipeline_layout, 0, - descriptor_set, {}); - cmdbuf.Draw(4, 1, 0, 0); + for (size_t i = 0; i < layer_count; i++) { + cmdbuf.PushConstants(graphics_pipeline_layout, VK_SHADER_STAGE_VERTEX_BIT, + push_constants[i]); + cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, graphics_pipeline_layout, 0, + descriptor_sets[i], {}); + cmdbuf.Draw(4, 1, 0, 0); + } + cmdbuf.EndRenderPass(); }); } -VkRenderPass WindowAdaptPass::GetRenderPass() { - return *render_pass; +VkDescriptorSetLayout WindowAdaptPass::GetDescriptorSetLayout() { + return *descriptor_set_layout; } -void WindowAdaptPass::CreateDescriptorPool(size_t num_images) { - const std::array pool_sizes{{ - { - .type = VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER, - .descriptorCount = static_cast(num_images), - }, - { - .type = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, - .descriptorCount = static_cast(num_images), - }, - }}; - - const VkDescriptorPoolCreateInfo ci{ - .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .maxSets = static_cast(num_images), - .poolSizeCount = static_cast(pool_sizes.size()), - .pPoolSizes = pool_sizes.data(), - }; - descriptor_pool = device.GetLogical().CreateDescriptorPool(ci); +VkRenderPass WindowAdaptPass::GetRenderPass() { + return *render_pass; } void WindowAdaptPass::CreateDescriptorSetLayout() { - const std::array layout_bindings{{ - { - .binding = 0, - .descriptorType = VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER, - .descriptorCount = 1, - .stageFlags = VK_SHADER_STAGE_VERTEX_BIT, - .pImmutableSamplers = nullptr, - }, - { - .binding = 1, - .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, - .descriptorCount = 1, - .stageFlags = VK_SHADER_STAGE_FRAGMENT_BIT, - .pImmutableSamplers = nullptr, - }, - }}; - - const VkDescriptorSetLayoutCreateInfo ci{ - .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .bindingCount = static_cast(layout_bindings.size()), - .pBindings = layout_bindings.data(), - }; - - descriptor_set_layout = device.GetLogical().CreateDescriptorSetLayout(ci); -} - -void WindowAdaptPass::CreateDescriptorSets(size_t num_images) { - const std::vector layouts(num_images, *descriptor_set_layout); - descriptor_sets = CreateWrappedDescriptorSets(descriptor_pool, layouts); -} - -void WindowAdaptPass::CreateBuffer(const MemoryAllocator& memory_allocator) { - const VkBufferCreateInfo ci{ - .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .size = sizeof(BufferData), - .usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT | - VK_BUFFER_USAGE_VERTEX_BUFFER_BIT | VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT, - .sharingMode = VK_SHARING_MODE_EXCLUSIVE, - .queueFamilyIndexCount = 0, - .pQueueFamilyIndices = nullptr, - }; - - buffer = memory_allocator.CreateBuffer(ci, MemoryUsage::Upload); + descriptor_set_layout = + CreateWrappedDescriptorSetLayout(device, {VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER}); } -void WindowAdaptPass::CreateRenderPass(VkFormat frame_format) { - const VkAttachmentDescription color_attachment{ - .flags = 0, - .format = frame_format, - .samples = VK_SAMPLE_COUNT_1_BIT, - .loadOp = VK_ATTACHMENT_LOAD_OP_CLEAR, - .storeOp = VK_ATTACHMENT_STORE_OP_STORE, - .stencilLoadOp = VK_ATTACHMENT_LOAD_OP_DONT_CARE, - .stencilStoreOp = VK_ATTACHMENT_STORE_OP_DONT_CARE, - .initialLayout = VK_IMAGE_LAYOUT_UNDEFINED, - .finalLayout = VK_IMAGE_LAYOUT_GENERAL, - }; - - const VkAttachmentReference color_attachment_ref{ - .attachment = 0, - .layout = VK_IMAGE_LAYOUT_GENERAL, - }; - - const VkSubpassDescription subpass_description{ - .flags = 0, - .pipelineBindPoint = VK_PIPELINE_BIND_POINT_GRAPHICS, - .inputAttachmentCount = 0, - .pInputAttachments = nullptr, - .colorAttachmentCount = 1, - .pColorAttachments = &color_attachment_ref, - .pResolveAttachments = nullptr, - .pDepthStencilAttachment = nullptr, - .preserveAttachmentCount = 0, - .pPreserveAttachments = nullptr, - }; - - const VkSubpassDependency dependency{ - .srcSubpass = VK_SUBPASS_EXTERNAL, - .dstSubpass = 0, - .srcStageMask = VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT, - .dstStageMask = VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT, - .srcAccessMask = 0, - .dstAccessMask = VK_ACCESS_COLOR_ATTACHMENT_READ_BIT | VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT, - .dependencyFlags = 0, - }; - - const VkRenderPassCreateInfo renderpass_ci{ - .sType = VK_STRUCTURE_TYPE_RENDER_PASS_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .attachmentCount = 1, - .pAttachments = &color_attachment, - .subpassCount = 1, - .pSubpasses = &subpass_description, - .dependencyCount = 1, - .pDependencies = &dependency, +void WindowAdaptPass::CreatePipelineLayout() { + const VkPushConstantRange range{ + .stageFlags = VK_SHADER_STAGE_VERTEX_BIT, + .offset = 0, + .size = sizeof(PresentPushConstants), }; - render_pass = device.GetLogical().CreateRenderPass(renderpass_ci); -} - -void WindowAdaptPass::CreateVertexShader() { - vertex_shader = BuildShader(device, VULKAN_PRESENT_VERT_SPV); -} - -void WindowAdaptPass::CreatePipelineLayout() { - const VkPipelineLayoutCreateInfo ci{ + pipeline_layout = device.GetLogical().CreatePipelineLayout(VkPipelineLayoutCreateInfo{ .sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO, .pNext = nullptr, .flags = 0, .setLayoutCount = 1, .pSetLayouts = descriptor_set_layout.address(), - .pushConstantRangeCount = 0, - .pPushConstantRanges = nullptr, - }; - pipeline_layout = device.GetLogical().CreatePipelineLayout(ci); -} - -void WindowAdaptPass::SetUniformData(BufferData& data, - const Layout::FramebufferLayout& layout) const { - data.uniform.modelview_matrix = - MakeOrthographicMatrix(static_cast(layout.width), static_cast(layout.height)); -} - -void WindowAdaptPass::SetVertexData(BufferData& data, const Layout::FramebufferLayout& layout, - const Common::Rectangle& crop) const { - // Map the coordinates to the screen. - const auto& screen = layout.screen; - const auto x = static_cast(screen.left); - const auto y = static_cast(screen.top); - const auto w = static_cast(screen.GetWidth()); - const auto h = static_cast(screen.GetHeight()); - - data.vertices[0] = ScreenRectVertex(x, y, crop.left, crop.top); - data.vertices[1] = ScreenRectVertex(x + w, y, crop.right, crop.top); - data.vertices[2] = ScreenRectVertex(x, y + h, crop.left, crop.bottom); - data.vertices[3] = ScreenRectVertex(x + w, y + h, crop.right, crop.bottom); + .pushConstantRangeCount = 1, + .pPushConstantRanges = &range, + }); } -void WindowAdaptPass::UpdateDescriptorSet(size_t image_index, VkImageView image_view) { - const VkDescriptorBufferInfo buffer_info{ - .buffer = *buffer, - .offset = offsetof(BufferData, uniform), - .range = sizeof(BufferData::uniform), - }; - - const VkWriteDescriptorSet ubo_write{ - .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET, - .pNext = nullptr, - .dstSet = descriptor_sets[image_index], - .dstBinding = 0, - .dstArrayElement = 0, - .descriptorCount = 1, - .descriptorType = VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER, - .pImageInfo = nullptr, - .pBufferInfo = &buffer_info, - .pTexelBufferView = nullptr, - }; - - const VkDescriptorImageInfo image_info{ - .sampler = *sampler, - .imageView = image_view, - .imageLayout = VK_IMAGE_LAYOUT_GENERAL, - }; - - const VkWriteDescriptorSet sampler_write{ - .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET, - .pNext = nullptr, - .dstSet = descriptor_sets[image_index], - .dstBinding = 1, - .dstArrayElement = 0, - .descriptorCount = 1, - .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, - .pImageInfo = &image_info, - .pBufferInfo = nullptr, - .pTexelBufferView = nullptr, - }; - - device.GetLogical().UpdateDescriptorSets(std::array{ubo_write, sampler_write}, {}); +void WindowAdaptPass::CreateVertexShader() { + vertex_shader = BuildShader(device, VULKAN_PRESENT_VERT_SPV); } -void WindowAdaptPass::ConfigureLayout(size_t image_index, VkImageView image_view, - const Layout::FramebufferLayout& layout, - const Common::Rectangle& crop_rect) { - BufferData data; - SetUniformData(data, layout); - SetVertexData(data, layout, crop_rect); - - const std::span mapped_span = buffer.Mapped(); - std::memcpy(mapped_span.data(), &data, sizeof(data)); - - UpdateDescriptorSet(image_index, image_view); +void WindowAdaptPass::CreateRenderPass(VkFormat frame_format) { + render_pass = CreateWrappedRenderPass(device, frame_format, VK_IMAGE_LAYOUT_UNDEFINED); } void WindowAdaptPass::CreatePipeline() { - const std::array shader_stages{{ - { - .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .stage = VK_SHADER_STAGE_VERTEX_BIT, - .module = *vertex_shader, - .pName = "main", - .pSpecializationInfo = nullptr, - }, - { - .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .stage = VK_SHADER_STAGE_FRAGMENT_BIT, - .module = *fragment_shader, - .pName = "main", - .pSpecializationInfo = nullptr, - }, - }}; - - const auto vertex_binding_description = ScreenRectVertex::GetDescription(); - const auto vertex_attrs_description = ScreenRectVertex::GetAttributes(); - - const VkPipelineVertexInputStateCreateInfo vertex_input_ci{ - .sType = VK_STRUCTURE_TYPE_PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .vertexBindingDescriptionCount = 1, - .pVertexBindingDescriptions = &vertex_binding_description, - .vertexAttributeDescriptionCount = u32{vertex_attrs_description.size()}, - .pVertexAttributeDescriptions = vertex_attrs_description.data(), - }; - - const VkPipelineInputAssemblyStateCreateInfo input_assembly_ci{ - .sType = VK_STRUCTURE_TYPE_PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .topology = VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP, - .primitiveRestartEnable = VK_FALSE, - }; - - const VkPipelineViewportStateCreateInfo viewport_state_ci{ - .sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_STATE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .viewportCount = 1, - .pViewports = nullptr, - .scissorCount = 1, - .pScissors = nullptr, - }; - - const VkPipelineRasterizationStateCreateInfo rasterization_ci{ - .sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_STATE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .depthClampEnable = VK_FALSE, - .rasterizerDiscardEnable = VK_FALSE, - .polygonMode = VK_POLYGON_MODE_FILL, - .cullMode = VK_CULL_MODE_NONE, - .frontFace = VK_FRONT_FACE_CLOCKWISE, - .depthBiasEnable = VK_FALSE, - .depthBiasConstantFactor = 0.0f, - .depthBiasClamp = 0.0f, - .depthBiasSlopeFactor = 0.0f, - .lineWidth = 1.0f, - }; - - const VkPipelineMultisampleStateCreateInfo multisampling_ci{ - .sType = VK_STRUCTURE_TYPE_PIPELINE_MULTISAMPLE_STATE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .rasterizationSamples = VK_SAMPLE_COUNT_1_BIT, - .sampleShadingEnable = VK_FALSE, - .minSampleShading = 0.0f, - .pSampleMask = nullptr, - .alphaToCoverageEnable = VK_FALSE, - .alphaToOneEnable = VK_FALSE, - }; - - const VkPipelineColorBlendAttachmentState color_blend_attachment{ - .blendEnable = VK_FALSE, - .srcColorBlendFactor = VK_BLEND_FACTOR_ZERO, - .dstColorBlendFactor = VK_BLEND_FACTOR_ZERO, - .colorBlendOp = VK_BLEND_OP_ADD, - .srcAlphaBlendFactor = VK_BLEND_FACTOR_ZERO, - .dstAlphaBlendFactor = VK_BLEND_FACTOR_ZERO, - .alphaBlendOp = VK_BLEND_OP_ADD, - .colorWriteMask = VK_COLOR_COMPONENT_R_BIT | VK_COLOR_COMPONENT_G_BIT | - VK_COLOR_COMPONENT_B_BIT | VK_COLOR_COMPONENT_A_BIT, - }; - - const VkPipelineColorBlendStateCreateInfo color_blend_ci{ - .sType = VK_STRUCTURE_TYPE_PIPELINE_COLOR_BLEND_STATE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .logicOpEnable = VK_FALSE, - .logicOp = VK_LOGIC_OP_COPY, - .attachmentCount = 1, - .pAttachments = &color_blend_attachment, - .blendConstants = {0.0f, 0.0f, 0.0f, 0.0f}, - }; - - static constexpr std::array dynamic_states{ - VK_DYNAMIC_STATE_VIEWPORT, - VK_DYNAMIC_STATE_SCISSOR, - }; - const VkPipelineDynamicStateCreateInfo dynamic_state_ci{ - .sType = VK_STRUCTURE_TYPE_PIPELINE_DYNAMIC_STATE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .dynamicStateCount = static_cast(dynamic_states.size()), - .pDynamicStates = dynamic_states.data(), - }; - - const VkGraphicsPipelineCreateInfo pipeline_ci{ - .sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO, - .pNext = nullptr, - .flags = 0, - .stageCount = static_cast(shader_stages.size()), - .pStages = shader_stages.data(), - .pVertexInputState = &vertex_input_ci, - .pInputAssemblyState = &input_assembly_ci, - .pTessellationState = nullptr, - .pViewportState = &viewport_state_ci, - .pRasterizationState = &rasterization_ci, - .pMultisampleState = &multisampling_ci, - .pDepthStencilState = nullptr, - .pColorBlendState = &color_blend_ci, - .pDynamicState = &dynamic_state_ci, - .layout = *pipeline_layout, - .renderPass = *render_pass, - .subpass = 0, - .basePipelineHandle = 0, - .basePipelineIndex = 0, - }; - - pipeline = device.GetLogical().CreateGraphicsPipeline(pipeline_ci); + pipeline = CreateWrappedPipeline(device, render_pass, pipeline_layout, + std::tie(vertex_shader, fragment_shader), false); } } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/window_adapt_pass.h b/src/video_core/renderer_vulkan/present/window_adapt_pass.h index 5309233a2..0e2edfc31 100644 --- a/src/video_core/renderer_vulkan/present/window_adapt_pass.h +++ b/src/video_core/renderer_vulkan/present/window_adapt_pass.h @@ -3,6 +3,8 @@ #pragma once +#include + #include "common/math_util.h" #include "video_core/vulkan_common/vulkan_wrapper.h" @@ -18,54 +20,39 @@ namespace Vulkan { class Device; struct Frame; -class MemoryAllocator; +class Layer; class Scheduler; +class RasterizerVulkan; class WindowAdaptPass final { public: - explicit WindowAdaptPass(const Device& device, const MemoryAllocator& memory_allocator, - size_t num_images, VkFormat frame_format, vk::Sampler&& sampler, + explicit WindowAdaptPass(const Device& device, VkFormat frame_format, vk::Sampler&& sampler, vk::ShaderModule&& fragment_shader); ~WindowAdaptPass(); - void Draw(Scheduler& scheduler, size_t image_index, VkImageView src_image_view, - VkExtent2D src_image_extent, const Common::Rectangle& crop_rect, + void Draw(RasterizerVulkan& rasterizer, Scheduler& scheduler, size_t image_index, + std::list& layers, std::span configs, const Layout::FramebufferLayout& layout, Frame* dst); + VkDescriptorSetLayout GetDescriptorSetLayout(); VkRenderPass GetRenderPass(); private: - struct BufferData; - - void SetUniformData(BufferData& data, const Layout::FramebufferLayout& layout) const; - void SetVertexData(BufferData& data, const Layout::FramebufferLayout& layout, - const Common::Rectangle& crop_rect) const; - void UpdateDescriptorSet(size_t image_index, VkImageView image_view); - void ConfigureLayout(size_t image_index, VkImageView image_view, - const Layout::FramebufferLayout& layout, - const Common::Rectangle& crop_rect); - - void CreateDescriptorPool(size_t num_images); void CreateDescriptorSetLayout(); - void CreateDescriptorSets(size_t num_images); void CreatePipelineLayout(); void CreateVertexShader(); void CreateRenderPass(VkFormat frame_format); void CreatePipeline(); - void CreateBuffer(const MemoryAllocator& memory_allocator); private: const Device& device; - vk::DescriptorPool descriptor_pool; vk::DescriptorSetLayout descriptor_set_layout; - vk::DescriptorSets descriptor_sets; vk::PipelineLayout pipeline_layout; vk::Sampler sampler; vk::ShaderModule vertex_shader; vk::ShaderModule fragment_shader; vk::RenderPass render_pass; vk::Pipeline pipeline; - vk::Buffer buffer; }; } // namespace Vulkan -- cgit v1.2.3 From a595e9e8a7a6a742481b1cd05455d3c639095413 Mon Sep 17 00:00:00 2001 From: Liam Date: Mon, 22 Jan 2024 12:40:50 -0500 Subject: nvnflinger/gpu: implement layer stack composition --- src/video_core/renderer_vulkan/present/util.cpp | 78 +++++++++++++++++++++++++ src/video_core/renderer_vulkan/present/util.h | 4 ++ 2 files changed, 82 insertions(+) (limited to 'src/video_core/renderer_vulkan/present') diff --git a/src/video_core/renderer_vulkan/present/util.cpp b/src/video_core/renderer_vulkan/present/util.cpp index 7bff1c436..6ee16595d 100644 --- a/src/video_core/renderer_vulkan/present/util.cpp +++ b/src/video_core/renderer_vulkan/present/util.cpp @@ -7,6 +7,20 @@ namespace Vulkan { +vk::Buffer CreateWrappedBuffer(MemoryAllocator& allocator, VkDeviceSize size, MemoryUsage usage) { + const VkBufferCreateInfo dst_buffer_info{ + .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO, + .pNext = nullptr, + .flags = 0, + .size = size, + .usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT, + .sharingMode = VK_SHARING_MODE_EXCLUSIVE, + .queueFamilyIndexCount = 0, + .pQueueFamilyIndices = nullptr, + }; + return allocator.CreateBuffer(dst_buffer_info, usage); +} + vk::Image CreateWrappedImage(MemoryAllocator& allocator, VkExtent2D dimensions, VkFormat format) { const VkImageCreateInfo image_ci{ .sType = VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO, @@ -96,6 +110,70 @@ void UploadImage(const Device& device, MemoryAllocator& allocator, Scheduler& sc scheduler.Finish(); } +void DownloadColorImage(vk::CommandBuffer& cmdbuf, VkImage image, VkBuffer buffer, + VkExtent3D extent) { + const VkImageMemoryBarrier read_barrier{ + .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER, + .pNext = nullptr, + .srcAccessMask = VK_ACCESS_MEMORY_WRITE_BIT, + .dstAccessMask = VK_ACCESS_TRANSFER_READ_BIT, + .oldLayout = VK_IMAGE_LAYOUT_GENERAL, + .newLayout = VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL, + .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, + .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, + .image = image, + .subresourceRange{ + .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .baseMipLevel = 0, + .levelCount = VK_REMAINING_MIP_LEVELS, + .baseArrayLayer = 0, + .layerCount = VK_REMAINING_ARRAY_LAYERS, + }, + }; + const VkImageMemoryBarrier image_write_barrier{ + .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER, + .pNext = nullptr, + .srcAccessMask = 0, + .dstAccessMask = VK_ACCESS_MEMORY_WRITE_BIT, + .oldLayout = VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL, + .newLayout = VK_IMAGE_LAYOUT_GENERAL, + .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, + .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED, + .image = image, + .subresourceRange{ + .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .baseMipLevel = 0, + .levelCount = VK_REMAINING_MIP_LEVELS, + .baseArrayLayer = 0, + .layerCount = VK_REMAINING_ARRAY_LAYERS, + }, + }; + static constexpr VkMemoryBarrier memory_write_barrier{ + .sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER, + .pNext = nullptr, + .srcAccessMask = VK_ACCESS_MEMORY_WRITE_BIT, + .dstAccessMask = VK_ACCESS_MEMORY_READ_BIT | VK_ACCESS_MEMORY_WRITE_BIT, + }; + const VkBufferImageCopy copy{ + .bufferOffset = 0, + .bufferRowLength = 0, + .bufferImageHeight = 0, + .imageSubresource{ + .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .mipLevel = 0, + .baseArrayLayer = 0, + .layerCount = 1, + }, + .imageOffset{.x = 0, .y = 0, .z = 0}, + .imageExtent{extent}, + }; + cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT, 0, + read_barrier); + cmdbuf.CopyImageToBuffer(image, VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL, buffer, copy); + cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, 0, + memory_write_barrier, nullptr, image_write_barrier); +} + vk::ImageView CreateWrappedImageView(const Device& device, vk::Image& image, VkFormat format) { return device.GetLogical().CreateImageView(VkImageViewCreateInfo{ .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO, diff --git a/src/video_core/renderer_vulkan/present/util.h b/src/video_core/renderer_vulkan/present/util.h index fb4e4a8e4..1104aaa15 100644 --- a/src/video_core/renderer_vulkan/present/util.h +++ b/src/video_core/renderer_vulkan/present/util.h @@ -11,12 +11,16 @@ namespace Vulkan { #define ARRAY_TO_SPAN(a) std::span(a, (sizeof(a) / sizeof(a[0]))) +vk::Buffer CreateWrappedBuffer(MemoryAllocator& allocator, VkDeviceSize size, MemoryUsage usage); + vk::Image CreateWrappedImage(MemoryAllocator& allocator, VkExtent2D dimensions, VkFormat format); void TransitionImageLayout(vk::CommandBuffer& cmdbuf, VkImage image, VkImageLayout target_layout, VkImageLayout source_layout = VK_IMAGE_LAYOUT_GENERAL); void UploadImage(const Device& device, MemoryAllocator& allocator, Scheduler& scheduler, vk::Image& image, VkExtent2D dimensions, VkFormat format, std::span initial_contents = {}); +void DownloadColorImage(vk::CommandBuffer& cmdbuf, VkImage image, VkBuffer buffer, + VkExtent3D extent); void ClearColorImage(vk::CommandBuffer& cmdbuf, VkImage image); vk::ImageView CreateWrappedImageView(const Device& device, vk::Image& image, VkFormat format); -- cgit v1.2.3