diff options
Diffstat (limited to 'src/video_core')
-rw-r--r-- | src/video_core/buffer_cache/buffer_cache.h | 29 | ||||
-rw-r--r-- | src/video_core/buffer_cache/buffer_cache_base.h | 3 | ||||
-rw-r--r-- | src/video_core/host1x/codecs/h264.cpp | 4 | ||||
-rw-r--r-- | src/video_core/renderer_vulkan/vk_pipeline_cache.cpp | 3 | ||||
-rw-r--r-- | src/video_core/renderer_vulkan/vk_rasterizer.cpp | 25 | ||||
-rw-r--r-- | src/video_core/vulkan_common/vulkan_device.cpp | 2 | ||||
-rw-r--r-- | src/video_core/vulkan_common/vulkan_memory_allocator.cpp | 2 |
7 files changed, 17 insertions, 51 deletions
diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h index fff57ffa9..98756e4da 100644 --- a/src/video_core/buffer_cache/buffer_cache.h +++ b/src/video_core/buffer_cache/buffer_cache.h @@ -131,33 +131,15 @@ std::optional<VideoCore::RasterizerDownloadArea> BufferCache<P>::GetFlushArea(VA template <class P> void BufferCache<P>::DownloadMemory(VAddr cpu_addr, u64 size) { - WaitOnAsyncFlushes(cpu_addr, size); ForEachBufferInRange(cpu_addr, size, [&](BufferId, Buffer& buffer) { DownloadBufferMemory(buffer, cpu_addr, size); }); } template <class P> -void BufferCache<P>::WaitOnAsyncFlushes(VAddr cpu_addr, u64 size) { - bool must_wait = false; - ForEachInOverlapCounter(async_downloads, cpu_addr, size, - [&](VAddr, VAddr, int) { must_wait = true; }); - bool must_release = false; - ForEachInRangeSet(pending_ranges, cpu_addr, size, [&](VAddr, VAddr) { must_release = true; }); - if (must_release) { - std::function<void()> tmp([]() {}); - rasterizer.SignalFence(std::move(tmp)); - } - if (must_wait || must_release) { - rasterizer.ReleaseFences(); - } -} - -template <class P> void BufferCache<P>::ClearDownload(IntervalType subtract_interval) { RemoveEachInOverlapCounter(async_downloads, subtract_interval, -1024); uncommitted_ranges.subtract(subtract_interval); - pending_ranges.subtract(subtract_interval); for (auto& interval_set : committed_ranges) { interval_set.subtract(subtract_interval); } @@ -177,7 +159,6 @@ bool BufferCache<P>::DMACopy(GPUVAddr src_address, GPUVAddr dest_address, u64 am } const IntervalType subtract_interval{*cpu_dest_address, *cpu_dest_address + amount}; - WaitOnAsyncFlushes(*cpu_src_address, static_cast<u32>(amount)); ClearDownload(subtract_interval); BufferId buffer_a; @@ -205,7 +186,6 @@ bool BufferCache<P>::DMACopy(GPUVAddr src_address, GPUVAddr dest_address, u64 am const IntervalType add_interval{new_base_address, new_base_address + size}; tmp_intervals.push_back(add_interval); uncommitted_ranges.add(add_interval); - pending_ranges.add(add_interval); }; ForEachInRangeSet(common_ranges, *cpu_src_address, amount, mirror); // This subtraction in this order is important for overlapping copies. @@ -492,7 +472,6 @@ void BufferCache<P>::CommitAsyncFlushesHigh() { } MICROPROFILE_SCOPE(GPU_DownloadMemory); - pending_ranges.clear(); auto it = committed_ranges.begin(); while (it != committed_ranges.end()) { auto& current_intervals = *it; @@ -1232,7 +1211,6 @@ void BufferCache<P>::MarkWrittenBuffer(BufferId buffer_id, VAddr cpu_addr, u32 s const IntervalType base_interval{cpu_addr, cpu_addr + size}; common_ranges.add(base_interval); uncommitted_ranges.add(base_interval); - pending_ranges.add(base_interval); } template <class P> @@ -1677,14 +1655,15 @@ typename BufferCache<P>::Binding BufferCache<P>::StorageBufferBinding(GPUVAddr s const bool is_nvn_cbuf = cbuf_index == 0; // The NVN driver buffer (index 0) is known to pack the SSBO address followed by its size. if (is_nvn_cbuf) { - return gpu_memory->Read<u32>(ssbo_addr + 8); + const u32 ssbo_size = gpu_memory->Read<u32>(ssbo_addr + 8); + if (ssbo_size != 0) { + return ssbo_size; + } } // Other titles (notably Doom Eternal) may use STG/LDG on buffer addresses in custom defined // cbufs, which do not store the sizes adjacent to the addresses, so use the fully // mapped buffer size for now. const u32 memory_layout_size = static_cast<u32>(gpu_memory->GetMemoryLayoutSize(gpu_addr)); - LOG_INFO(HW_GPU, "Binding storage buffer for cbuf index {}, MemoryLayoutSize 0x{:X}", - cbuf_index, memory_layout_size); return memory_layout_size; }(); const std::optional<VAddr> cpu_addr = gpu_memory->GpuToCpuAddress(gpu_addr); diff --git a/src/video_core/buffer_cache/buffer_cache_base.h b/src/video_core/buffer_cache/buffer_cache_base.h index 0445ec47f..ac00d4d9d 100644 --- a/src/video_core/buffer_cache/buffer_cache_base.h +++ b/src/video_core/buffer_cache/buffer_cache_base.h @@ -381,8 +381,6 @@ private: void RunGarbageCollector(); - void WaitOnAsyncFlushes(VAddr cpu_addr, u64 size); - void BindHostIndexBuffer(); void BindHostVertexBuffers(); @@ -547,7 +545,6 @@ private: IntervalSet uncommitted_ranges; IntervalSet common_ranges; IntervalSet cached_ranges; - IntervalSet pending_ranges; std::deque<IntervalSet> committed_ranges; // Async Buffers diff --git a/src/video_core/host1x/codecs/h264.cpp b/src/video_core/host1x/codecs/h264.cpp index e87bd65fa..6ce179167 100644 --- a/src/video_core/host1x/codecs/h264.cpp +++ b/src/video_core/host1x/codecs/h264.cpp @@ -111,7 +111,7 @@ const std::vector<u8>& H264::ComposeFrame(const Host1x::NvdecCommon::NvdecRegist writer.WriteUe(0); writer.WriteBit(context.h264_parameter_set.entropy_coding_mode_flag != 0); - writer.WriteBit(false); + writer.WriteBit(context.h264_parameter_set.pic_order_present_flag != 0); writer.WriteUe(0); writer.WriteUe(context.h264_parameter_set.num_refidx_l0_default_active); writer.WriteUe(context.h264_parameter_set.num_refidx_l1_default_active); @@ -129,7 +129,7 @@ const std::vector<u8>& H264::ComposeFrame(const Host1x::NvdecCommon::NvdecRegist writer.WriteBit(context.h264_parameter_set.redundant_pic_cnt_present_flag != 0); writer.WriteBit(context.h264_parameter_set.transform_8x8_mode_flag != 0); - writer.WriteBit(true); + writer.WriteBit(true); // pic_scaling_matrix_present_flag for (s32 index = 0; index < 6; index++) { writer.WriteBit(true); diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp index 596996bec..66dfe5733 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp @@ -698,7 +698,8 @@ std::unique_ptr<ComputePipeline> PipelineCache::CreateComputePipeline( PipelineStatistics* statistics, bool build_in_parallel) try { // TODO: Remove this when Intel fixes their shader compiler. // https://github.com/IGCIT/Intel-GPU-Community-Issue-Tracker-IGCIT/issues/159 - if (device.GetDriverID() == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS) { + if (device.GetDriverID() == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS && + !Settings::values.enable_compute_pipelines.GetValue()) { LOG_ERROR(Render_Vulkan, "Skipping 0x{:016x}", key.Hash()); return nullptr; } diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index 64bd2f6a5..8d3a9736b 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -348,25 +348,12 @@ void RasterizerVulkan::Clear(u32 layer_count) { const u32 color_attachment = regs.clear_surface.RT; if (use_color && framebuffer->HasAspectColorBit(color_attachment)) { - VkClearValue clear_value; - bool is_integer = false; - bool is_signed = false; - size_t int_size = 8; - for (std::size_t i = 0; i < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; ++i) { - const auto& this_rt = regs.rt[i]; - if (this_rt.Address() == 0) { - continue; - } - if (this_rt.format == Tegra::RenderTargetFormat::NONE) { - continue; - } - const auto format = - VideoCore::Surface::PixelFormatFromRenderTargetFormat(this_rt.format); - is_integer = IsPixelFormatInteger(format); - is_signed = IsPixelFormatSignedInteger(format); - int_size = PixelComponentSizeBitsInteger(format); - break; - } + const auto format = + VideoCore::Surface::PixelFormatFromRenderTargetFormat(regs.rt[color_attachment].format); + bool is_integer = IsPixelFormatInteger(format); + bool is_signed = IsPixelFormatSignedInteger(format); + size_t int_size = PixelComponentSizeBitsInteger(format); + VkClearValue clear_value{}; if (!is_integer) { std::memcpy(clear_value.color.float32, regs.clear_color.data(), regs.clear_color.size() * sizeof(f32)); diff --git a/src/video_core/vulkan_common/vulkan_device.cpp b/src/video_core/vulkan_common/vulkan_device.cpp index 7132bae99..9a8763e6a 100644 --- a/src/video_core/vulkan_common/vulkan_device.cpp +++ b/src/video_core/vulkan_common/vulkan_device.cpp @@ -1017,6 +1017,8 @@ void Device::CollectPhysicalMemoryInfo() { device_access_memory += mem_properties.memoryHeaps[element].size; } if (!is_integrated) { + const u64 reserve_memory = std::min<u64>(device_access_memory / 8, 1_GiB); + device_access_memory -= reserve_memory; return; } const s64 available_memory = static_cast<s64>(device_access_memory - device_initial_usage); diff --git a/src/video_core/vulkan_common/vulkan_memory_allocator.cpp b/src/video_core/vulkan_common/vulkan_memory_allocator.cpp index 1732866e0..e28a556f8 100644 --- a/src/video_core/vulkan_common/vulkan_memory_allocator.cpp +++ b/src/video_core/vulkan_common/vulkan_memory_allocator.cpp @@ -147,7 +147,7 @@ public: /// Returns whether this allocation is compatible with the arguments. [[nodiscard]] bool IsCompatible(VkMemoryPropertyFlags flags, u32 type_mask) const { - return (flags & property_flags) == property_flags && (type_mask & shifted_memory_type) != 0; + return (flags & property_flags) == flags && (type_mask & shifted_memory_type) != 0; } private: |