diff options
Diffstat (limited to 'src/video_core/renderer_vulkan')
12 files changed, 92 insertions, 64 deletions
diff --git a/src/video_core/renderer_vulkan/blit_image.cpp b/src/video_core/renderer_vulkan/blit_image.cpp index 1032c9d12..f01d2394e 100644 --- a/src/video_core/renderer_vulkan/blit_image.cpp +++ b/src/video_core/renderer_vulkan/blit_image.cpp @@ -9,6 +9,7 @@ #include "video_core/host_shaders/blit_color_float_frag_spv.h" #include "video_core/host_shaders/convert_abgr8_to_d24s8_frag_spv.h" #include "video_core/host_shaders/convert_d24s8_to_abgr8_frag_spv.h" +#include "video_core/host_shaders/convert_d32f_to_abgr8_frag_spv.h" #include "video_core/host_shaders/convert_depth_to_float_frag_spv.h" #include "video_core/host_shaders/convert_float_to_depth_frag_spv.h" #include "video_core/host_shaders/convert_s8d24_to_abgr8_frag_spv.h" @@ -433,6 +434,7 @@ BlitImageHelper::BlitImageHelper(const Device& device_, Scheduler& scheduler_, convert_depth_to_float_frag(BuildShader(device, CONVERT_DEPTH_TO_FLOAT_FRAG_SPV)), convert_float_to_depth_frag(BuildShader(device, CONVERT_FLOAT_TO_DEPTH_FRAG_SPV)), convert_abgr8_to_d24s8_frag(BuildShader(device, CONVERT_ABGR8_TO_D24S8_FRAG_SPV)), + convert_d32f_to_abgr8_frag(BuildShader(device, CONVERT_D32F_TO_ABGR8_FRAG_SPV)), convert_d24s8_to_abgr8_frag(BuildShader(device, CONVERT_D24S8_TO_ABGR8_FRAG_SPV)), convert_s8d24_to_abgr8_frag(BuildShader(device, CONVERT_S8D24_TO_ABGR8_FRAG_SPV)), linear_sampler(device.GetLogical().CreateSampler(SAMPLER_CREATE_INFO<VK_FILTER_LINEAR>)), @@ -557,6 +559,13 @@ void BlitImageHelper::ConvertABGR8ToD24S8(const Framebuffer* dst_framebuffer, Convert(*convert_abgr8_to_d24s8_pipeline, dst_framebuffer, src_image_view); } +void BlitImageHelper::ConvertD32FToABGR8(const Framebuffer* dst_framebuffer, + ImageView& src_image_view) { + ConvertPipelineColorTargetEx(convert_d32f_to_abgr8_pipeline, dst_framebuffer->RenderPass(), + convert_d32f_to_abgr8_frag); + ConvertDepthStencil(*convert_d32f_to_abgr8_pipeline, dst_framebuffer, src_image_view); +} + void BlitImageHelper::ConvertD24S8ToABGR8(const Framebuffer* dst_framebuffer, ImageView& src_image_view) { ConvertPipelineColorTargetEx(convert_d24s8_to_abgr8_pipeline, dst_framebuffer->RenderPass(), @@ -609,6 +618,8 @@ void BlitImageHelper::ClearDepthStencil(const Framebuffer* dst_framebuffer, bool const VkPipelineLayout layout = *clear_color_pipeline_layout; scheduler.RequestRenderpass(dst_framebuffer); scheduler.Record([pipeline, layout, clear_depth, dst_region](vk::CommandBuffer cmdbuf) { + constexpr std::array blend_constants{0.0f, 0.0f, 0.0f, 0.0f}; + cmdbuf.SetBlendConstants(blend_constants.data()); cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline); BindBlitState(cmdbuf, dst_region); cmdbuf.PushConstants(layout, VK_SHADER_STAGE_FRAGMENT_BIT, clear_depth); @@ -865,7 +876,7 @@ VkPipeline BlitImageHelper::FindOrEmplaceClearStencilPipeline( .sType = VK_STRUCTURE_TYPE_PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO, .pNext = nullptr, .flags = 0, - .depthTestEnable = VK_FALSE, + .depthTestEnable = key.depth_clear, .depthWriteEnable = key.depth_clear, .depthCompareOp = VK_COMPARE_OP_ALWAYS, .depthBoundsTestEnable = VK_FALSE, diff --git a/src/video_core/renderer_vulkan/blit_image.h b/src/video_core/renderer_vulkan/blit_image.h index dcfe217aa..a032c71fb 100644 --- a/src/video_core/renderer_vulkan/blit_image.h +++ b/src/video_core/renderer_vulkan/blit_image.h @@ -67,6 +67,8 @@ public: void ConvertABGR8ToD24S8(const Framebuffer* dst_framebuffer, const ImageView& src_image_view); + void ConvertD32FToABGR8(const Framebuffer* dst_framebuffer, ImageView& src_image_view); + void ConvertD24S8ToABGR8(const Framebuffer* dst_framebuffer, ImageView& src_image_view); void ConvertS8D24ToABGR8(const Framebuffer* dst_framebuffer, ImageView& src_image_view); @@ -128,6 +130,7 @@ private: vk::ShaderModule convert_depth_to_float_frag; vk::ShaderModule convert_float_to_depth_frag; vk::ShaderModule convert_abgr8_to_d24s8_frag; + vk::ShaderModule convert_d32f_to_abgr8_frag; vk::ShaderModule convert_d24s8_to_abgr8_frag; vk::ShaderModule convert_s8d24_to_abgr8_frag; vk::Sampler linear_sampler; @@ -146,6 +149,7 @@ private: vk::Pipeline convert_d16_to_r16_pipeline; vk::Pipeline convert_r16_to_d16_pipeline; vk::Pipeline convert_abgr8_to_d24s8_pipeline; + vk::Pipeline convert_d32f_to_abgr8_pipeline; vk::Pipeline convert_d24s8_to_abgr8_pipeline; vk::Pipeline convert_s8d24_to_abgr8_pipeline; }; diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp index 208e88533..a08f2f67f 100644 --- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp +++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp @@ -214,8 +214,9 @@ struct FormatTuple { {VK_FORMAT_E5B9G9R9_UFLOAT_PACK32}, // E5B9G9R9_FLOAT // Depth formats - {VK_FORMAT_D32_SFLOAT, Attachable}, // D32_FLOAT - {VK_FORMAT_D16_UNORM, Attachable}, // D16_UNORM + {VK_FORMAT_D32_SFLOAT, Attachable}, // D32_FLOAT + {VK_FORMAT_D16_UNORM, Attachable}, // D16_UNORM + {VK_FORMAT_X8_D24_UNORM_PACK32, Attachable}, // X8_D24_UNORM // Stencil formats {VK_FORMAT_S8_UINT, Attachable}, // S8_UINT diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.h b/src/video_core/renderer_vulkan/renderer_vulkan.h index 590bc1c64..14e257cf7 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.h +++ b/src/video_core/renderer_vulkan/renderer_vulkan.h @@ -56,10 +56,6 @@ public: return device.GetDriverName(); } - void NotifySurfaceChanged() override { - present_manager.NotifySurfaceChanged(); - } - private: void Report() const; diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.cpp b/src/video_core/renderer_vulkan/vk_blit_screen.cpp index 31928bb94..52fc142d1 100644 --- a/src/video_core/renderer_vulkan/vk_blit_screen.cpp +++ b/src/video_core/renderer_vulkan/vk_blit_screen.cpp @@ -96,6 +96,7 @@ std::size_t GetSizeInBytes(const Tegra::FramebufferConfig& 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; diff --git a/src/video_core/renderer_vulkan/vk_present_manager.cpp b/src/video_core/renderer_vulkan/vk_present_manager.cpp index d681bd22a..2ef36583b 100644 --- a/src/video_core/renderer_vulkan/vk_present_manager.cpp +++ b/src/video_core/renderer_vulkan/vk_present_manager.cpp @@ -103,8 +103,7 @@ PresentManager::PresentManager(const vk::Instance& instance_, surface{surface_}, blit_supported{CanBlitToSwapchain(device.GetPhysical(), swapchain.GetImageViewFormat())}, use_present_thread{Settings::values.async_presentation.GetValue()}, - image_count{swapchain.GetImageCount()}, last_render_surface{ - render_window_.GetWindowInfo().render_surface} { + image_count{swapchain.GetImageCount()} { auto& dld = device.GetLogical(); cmdpool = dld.CreateCommandPool({ @@ -289,44 +288,36 @@ void PresentManager::PresentThread(std::stop_token token) { } } -void PresentManager::NotifySurfaceChanged() { -#ifdef ANDROID - std::scoped_lock lock{recreate_surface_mutex}; - recreate_surface_cv.notify_one(); -#endif +void PresentManager::RecreateSwapchain(Frame* frame) { + swapchain.Create(*surface, frame->width, frame->height, frame->is_srgb); + image_count = swapchain.GetImageCount(); } void PresentManager::CopyToSwapchain(Frame* frame) { - MICROPROFILE_SCOPE(Vulkan_CopyToSwapchain); - - const auto recreate_swapchain = [&] { - swapchain.Create(*surface, frame->width, frame->height, frame->is_srgb); - image_count = swapchain.GetImageCount(); - }; - -#ifdef ANDROID - std::unique_lock lock{recreate_surface_mutex}; - - const auto needs_recreation = [&] { - if (last_render_surface != render_window.GetWindowInfo().render_surface) { - return true; - } - if (swapchain.NeedsRecreation(frame->is_srgb)) { - return true; + bool requires_recreation = false; + + while (true) { + try { + // Recreate surface and swapchain if needed. + if (requires_recreation) { + surface = CreateSurface(instance, render_window.GetWindowInfo()); + RecreateSwapchain(frame); + } + + // Draw to swapchain. + return CopyToSwapchainImpl(frame); + } catch (const vk::Exception& except) { + if (except.GetResult() != VK_ERROR_SURFACE_LOST_KHR) { + throw; + } + + requires_recreation = true; } - return false; - }; - - recreate_surface_cv.wait_for(lock, std::chrono::milliseconds(400), - [&]() { return !needs_recreation(); }); - - // If the frontend recreated the surface, recreate the renderer surface and swapchain. - if (last_render_surface != render_window.GetWindowInfo().render_surface) { - last_render_surface = render_window.GetWindowInfo().render_surface; - surface = CreateSurface(instance, render_window.GetWindowInfo()); - recreate_swapchain(); } -#endif +} + +void PresentManager::CopyToSwapchainImpl(Frame* frame) { + MICROPROFILE_SCOPE(Vulkan_CopyToSwapchain); // If the size or colorspace of the incoming frames has changed, recreate the swapchain // to account for that. @@ -334,11 +325,11 @@ void PresentManager::CopyToSwapchain(Frame* frame) { const bool size_changed = swapchain.GetWidth() != frame->width || swapchain.GetHeight() != frame->height; if (srgb_changed || size_changed) { - recreate_swapchain(); + RecreateSwapchain(frame); } while (swapchain.AcquireNextImage()) { - recreate_swapchain(); + RecreateSwapchain(frame); } const vk::CommandBuffer cmdbuf{frame->cmdbuf}; @@ -488,4 +479,4 @@ void PresentManager::CopyToSwapchain(Frame* frame) { swapchain.Present(render_semaphore); } -} // namespace Vulkan
\ No newline at end of file +} // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/vk_present_manager.h b/src/video_core/renderer_vulkan/vk_present_manager.h index 83e859416..a3d825fe6 100644 --- a/src/video_core/renderer_vulkan/vk_present_manager.h +++ b/src/video_core/renderer_vulkan/vk_present_manager.h @@ -54,14 +54,15 @@ public: /// Waits for the present thread to finish presenting all queued frames. void WaitPresent(); - /// This is called to notify the rendering backend of a surface change - void NotifySurfaceChanged(); - private: void PresentThread(std::stop_token token); void CopyToSwapchain(Frame* frame); + void CopyToSwapchainImpl(Frame* frame); + + void RecreateSwapchain(Frame* frame); + private: const vk::Instance& instance; Core::Frontend::EmuWindow& render_window; @@ -76,16 +77,13 @@ private: std::queue<Frame*> free_queue; std::condition_variable_any frame_cv; std::condition_variable free_cv; - std::condition_variable recreate_surface_cv; std::mutex swapchain_mutex; - std::mutex recreate_surface_mutex; std::mutex queue_mutex; std::mutex free_mutex; std::jthread present_thread; bool blit_supported; bool use_present_thread; std::size_t image_count{}; - void* last_render_surface{}; }; } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/vk_query_cache.cpp b/src/video_core/renderer_vulkan/vk_query_cache.cpp index a32da3ba3..2edaafa7e 100644 --- a/src/video_core/renderer_vulkan/vk_query_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_query_cache.cpp @@ -506,6 +506,7 @@ private: SetAccumulationValue(query->value); Free(index); }); + rasterizer->SyncOperation(std::move(func)); } template <bool is_resolve> @@ -1160,7 +1161,7 @@ struct QueryCacheRuntimeImpl { cpu_memory_), primitives_needed_minus_suceeded_streamer( static_cast<size_t>(QueryType::StreamingPrimitivesNeededMinusSucceeded), runtime, 0u), - hcr_setup{}, hcr_is_set{}, is_hcr_running{} { + hcr_setup{}, hcr_is_set{}, is_hcr_running{}, maxwell3d{} { hcr_setup.sType = VK_STRUCTURE_TYPE_CONDITIONAL_RENDERING_BEGIN_INFO_EXT; hcr_setup.pNext = nullptr; @@ -1235,7 +1236,9 @@ void QueryCacheRuntime::Bind3DEngine(Maxwell3D* maxwell3d) { template <typename Func> void QueryCacheRuntime::View3DRegs(Func&& func) { - func(*impl->maxwell3d); + if (impl->maxwell3d) { + func(*impl->maxwell3d); + } } void QueryCacheRuntime::EndHostConditionalRendering() { diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index 1628d76d6..83f2b6045 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -422,7 +422,8 @@ void RasterizerVulkan::Clear(u32 layer_count) { return; } - if (use_stencil && regs.stencil_front_mask != 0xFF && regs.stencil_front_mask != 0) { + if (use_stencil && framebuffer->HasAspectStencilBit() && regs.stencil_front_mask != 0xFF && + regs.stencil_front_mask != 0) { Region2D dst_region = { Offset2D{.x = clear_rect.rect.offset.x, .y = clear_rect.rect.offset.y}, Offset2D{.x = clear_rect.rect.offset.x + static_cast<s32>(clear_rect.rect.extent.width), diff --git a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp index ce92f66ab..b278614e6 100644 --- a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp +++ b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp @@ -24,25 +24,38 @@ using namespace Common::Literals; // Maximum potential alignment of a Vulkan buffer constexpr VkDeviceSize MAX_ALIGNMENT = 256; -// Maximum size to put elements in the stream buffer -constexpr VkDeviceSize MAX_STREAM_BUFFER_REQUEST_SIZE = 8_MiB; // Stream buffer size in bytes -constexpr VkDeviceSize STREAM_BUFFER_SIZE = 128_MiB; -constexpr VkDeviceSize REGION_SIZE = STREAM_BUFFER_SIZE / StagingBufferPool::NUM_SYNCS; +constexpr VkDeviceSize MAX_STREAM_BUFFER_SIZE = 128_MiB; -size_t Region(size_t iterator) noexcept { - return iterator / REGION_SIZE; +size_t GetStreamBufferSize(const Device& device) { + VkDeviceSize size{0}; + if (device.HasDebuggingToolAttached()) { + ForEachDeviceLocalHostVisibleHeap(device, [&size](size_t index, VkMemoryHeap& heap) { + size = std::max(size, heap.size); + }); + // If rebar is not supported, cut the max heap size to 40%. This will allow 2 captures to be + // loaded at the same time in RenderDoc. If rebar is supported, this shouldn't be an issue + // as the heap will be much larger. + if (size <= 256_MiB) { + size = size * 40 / 100; + } + } else { + size = MAX_STREAM_BUFFER_SIZE; + } + return std::min(Common::AlignUp(size, MAX_ALIGNMENT), MAX_STREAM_BUFFER_SIZE); } } // Anonymous namespace StagingBufferPool::StagingBufferPool(const Device& device_, MemoryAllocator& memory_allocator_, Scheduler& scheduler_) - : device{device_}, memory_allocator{memory_allocator_}, scheduler{scheduler_} { + : device{device_}, memory_allocator{memory_allocator_}, scheduler{scheduler_}, + stream_buffer_size{GetStreamBufferSize(device)}, region_size{stream_buffer_size / + StagingBufferPool::NUM_SYNCS} { VkBufferCreateInfo stream_ci = { .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO, .pNext = nullptr, .flags = 0, - .size = STREAM_BUFFER_SIZE, + .size = stream_buffer_size, .usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT | VK_BUFFER_USAGE_INDEX_BUFFER_BIT | VK_BUFFER_USAGE_STORAGE_BUFFER_BIT, .sharingMode = VK_SHARING_MODE_EXCLUSIVE, @@ -63,7 +76,7 @@ StagingBufferPool::StagingBufferPool(const Device& device_, MemoryAllocator& mem StagingBufferPool::~StagingBufferPool() = default; StagingBufferRef StagingBufferPool::Request(size_t size, MemoryUsage usage, bool deferred) { - if (!deferred && usage == MemoryUsage::Upload && size <= MAX_STREAM_BUFFER_REQUEST_SIZE) { + if (!deferred && usage == MemoryUsage::Upload && size <= region_size) { return GetStreamBuffer(size); } return GetStagingBuffer(size, usage, deferred); @@ -101,7 +114,7 @@ StagingBufferRef StagingBufferPool::GetStreamBuffer(size_t size) { used_iterator = iterator; free_iterator = std::max(free_iterator, iterator + size); - if (iterator + size >= STREAM_BUFFER_SIZE) { + if (iterator + size >= stream_buffer_size) { std::fill(sync_ticks.begin() + Region(used_iterator), sync_ticks.begin() + NUM_SYNCS, current_tick); used_iterator = 0; diff --git a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.h b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.h index 5f69f08b1..d3deb9072 100644 --- a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.h +++ b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.h @@ -90,6 +90,9 @@ private: void ReleaseCache(MemoryUsage usage); void ReleaseLevel(StagingBuffersCache& cache, size_t log2); + size_t Region(size_t iter) const noexcept { + return iter / region_size; + } const Device& device; MemoryAllocator& memory_allocator; @@ -97,6 +100,8 @@ private: vk::Buffer stream_buffer; std::span<u8> stream_pointer; + VkDeviceSize stream_buffer_size; + VkDeviceSize region_size; size_t iterator = 0; size_t used_iterator = 0; diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.cpp b/src/video_core/renderer_vulkan/vk_texture_cache.cpp index 71fdec809..00ab47268 100644 --- a/src/video_core/renderer_vulkan/vk_texture_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_texture_cache.cpp @@ -238,6 +238,7 @@ constexpr VkBorderColor ConvertBorderColor(const std::array<float, 4>& color) { return any_r ? VK_IMAGE_ASPECT_STENCIL_BIT : VK_IMAGE_ASPECT_DEPTH_BIT; case PixelFormat::D16_UNORM: case PixelFormat::D32_FLOAT: + case PixelFormat::X8_D24_UNORM: return VK_IMAGE_ASPECT_DEPTH_BIT; case PixelFormat::S8_UINT: return VK_IMAGE_ASPECT_STENCIL_BIT; @@ -1200,6 +1201,9 @@ void TextureCacheRuntime::ConvertImage(Framebuffer* dst, ImageView& dst_view, Im if (src_view.format == PixelFormat::D24_UNORM_S8_UINT) { return blit_image_helper.ConvertS8D24ToABGR8(dst, src_view); } + if (src_view.format == PixelFormat::D32_FLOAT) { + return blit_image_helper.ConvertD32FToABGR8(dst, src_view); + } break; case PixelFormat::R32_FLOAT: if (src_view.format == PixelFormat::D32_FLOAT) { |