diff options
Diffstat (limited to 'src/video_core')
-rw-r--r-- | src/video_core/buffer_cache/buffer_cache_base.h | 4 | ||||
-rw-r--r-- | src/video_core/engines/draw_manager.cpp | 24 | ||||
-rw-r--r-- | src/video_core/engines/draw_manager.h | 2 | ||||
-rw-r--r-- | src/video_core/host1x/codecs/codec.cpp | 10 | ||||
-rw-r--r-- | src/video_core/host_shaders/convert_d24s8_to_abgr8.frag | 8 | ||||
-rw-r--r-- | src/video_core/host_shaders/convert_s8d24_to_abgr8.frag | 8 | ||||
-rw-r--r-- | src/video_core/renderer_vulkan/vk_query_cache.cpp | 1 | ||||
-rw-r--r-- | src/video_core/renderer_vulkan/vk_rasterizer.cpp | 13 | ||||
-rw-r--r-- | src/video_core/renderer_vulkan/vk_render_pass_cache.cpp | 2 | ||||
-rw-r--r-- | src/video_core/texture_cache/formatter.cpp | 8 | ||||
-rw-r--r-- | src/video_core/texture_cache/samples_helper.h | 2 | ||||
-rw-r--r-- | src/video_core/texture_cache/util.cpp | 11 |
12 files changed, 68 insertions, 25 deletions
diff --git a/src/video_core/buffer_cache/buffer_cache_base.h b/src/video_core/buffer_cache/buffer_cache_base.h index c4f6e8d12..eed267361 100644 --- a/src/video_core/buffer_cache/buffer_cache_base.h +++ b/src/video_core/buffer_cache/buffer_cache_base.h @@ -62,7 +62,11 @@ using BufferId = SlotId; using VideoCore::Surface::PixelFormat; using namespace Common::Literals; +#ifdef __APPLE__ +constexpr u32 NUM_VERTEX_BUFFERS = 16; +#else constexpr u32 NUM_VERTEX_BUFFERS = 32; +#endif constexpr u32 NUM_TRANSFORM_FEEDBACK_BUFFERS = 4; constexpr u32 NUM_GRAPHICS_UNIFORM_BUFFERS = 18; constexpr u32 NUM_COMPUTE_UNIFORM_BUFFERS = 8; diff --git a/src/video_core/engines/draw_manager.cpp b/src/video_core/engines/draw_manager.cpp index f34090791..d77ff455b 100644 --- a/src/video_core/engines/draw_manager.cpp +++ b/src/video_core/engines/draw_manager.cpp @@ -48,8 +48,14 @@ void DrawManager::ProcessMethodCall(u32 method, u32 argument) { SetInlineIndexBuffer(regs.inline_index_4x8.index3); break; case MAXWELL3D_REG_INDEX(vertex_array_instance_first): + DrawArrayInstanced(regs.vertex_array_instance_first.topology.Value(), + regs.vertex_array_instance_first.start.Value(), + regs.vertex_array_instance_first.count.Value(), false); + break; case MAXWELL3D_REG_INDEX(vertex_array_instance_subsequent): { - LOG_WARNING(HW_GPU, "(STUBBED) called"); + DrawArrayInstanced(regs.vertex_array_instance_subsequent.topology.Value(), + regs.vertex_array_instance_subsequent.start.Value(), + regs.vertex_array_instance_subsequent.count.Value(), true); break; } case MAXWELL3D_REG_INDEX(draw_texture.src_y0): { @@ -84,6 +90,22 @@ void DrawManager::DrawArray(PrimitiveTopology topology, u32 vertex_first, u32 ve ProcessDraw(false, num_instances); } +void DrawManager::DrawArrayInstanced(PrimitiveTopology topology, u32 vertex_first, u32 vertex_count, + bool subsequent) { + draw_state.topology = topology; + draw_state.vertex_buffer.first = vertex_first; + draw_state.vertex_buffer.count = vertex_count; + + if (!subsequent) { + draw_state.instance_count = 1; + } + + draw_state.base_instance = draw_state.instance_count - 1; + draw_state.draw_mode = DrawMode::Instance; + draw_state.instance_count++; + ProcessDraw(false, 1); +} + void DrawManager::DrawIndex(PrimitiveTopology topology, u32 index_first, u32 index_count, u32 base_index, u32 base_instance, u32 num_instances) { const auto& regs{maxwell3d->regs}; diff --git a/src/video_core/engines/draw_manager.h b/src/video_core/engines/draw_manager.h index 18d959143..cfc8127fc 100644 --- a/src/video_core/engines/draw_manager.h +++ b/src/video_core/engines/draw_manager.h @@ -66,6 +66,8 @@ public: void DrawArray(PrimitiveTopology topology, u32 vertex_first, u32 vertex_count, u32 base_instance, u32 num_instances); + void DrawArrayInstanced(PrimitiveTopology topology, u32 vertex_first, u32 vertex_count, + bool subsequent); void DrawIndex(PrimitiveTopology topology, u32 index_first, u32 index_count, u32 base_index, u32 base_instance, u32 num_instances); diff --git a/src/video_core/host1x/codecs/codec.cpp b/src/video_core/host1x/codecs/codec.cpp index 8d7da50fc..dbcf508e5 100644 --- a/src/video_core/host1x/codecs/codec.cpp +++ b/src/video_core/host1x/codecs/codec.cpp @@ -137,16 +137,6 @@ bool Codec::CreateGpuAvDevice() { break; } if ((config->methods & HW_CONFIG_METHOD) != 0 && config->device_type == type) { -#if defined(__unix__) - // Some linux decoding backends are reported to crash with this config method - // TODO(ameerj): Properly support this method - if ((config->methods & AV_CODEC_HW_CONFIG_METHOD_HW_FRAMES_CTX) != 0) { - // skip zero-copy decoders, we don't currently support them - LOG_DEBUG(Service_NVDRV, "Skipping decoder {} with unsupported capability {}.", - av_hwdevice_get_type_name(type), config->methods); - continue; - } -#endif LOG_INFO(Service_NVDRV, "Using {} GPU decoder", av_hwdevice_get_type_name(type)); av_codec_ctx->pix_fmt = config->pix_fmt; return true; diff --git a/src/video_core/host_shaders/convert_d24s8_to_abgr8.frag b/src/video_core/host_shaders/convert_d24s8_to_abgr8.frag index d33131d7c..b81a54056 100644 --- a/src/video_core/host_shaders/convert_d24s8_to_abgr8.frag +++ b/src/video_core/host_shaders/convert_d24s8_to_abgr8.frag @@ -3,16 +3,16 @@ #version 450 +precision mediump int; +precision highp float; + layout(binding = 0) uniform sampler2D depth_tex; -layout(binding = 1) uniform isampler2D stencil_tex; +layout(binding = 1) uniform usampler2D stencil_tex; layout(location = 0) out vec4 color; void main() { ivec2 coord = ivec2(gl_FragCoord.xy); - uint depth = uint(textureLod(depth_tex, coord, 0).r * (exp2(24.0) - 1.0f)); - uint stencil = uint(textureLod(stencil_tex, coord, 0).r); - highp uint depth_val = uint(textureLod(depth_tex, coord, 0).r * (exp2(32.0) - 1.0)); lowp uint stencil_val = textureLod(stencil_tex, coord, 0).r; diff --git a/src/video_core/host_shaders/convert_s8d24_to_abgr8.frag b/src/video_core/host_shaders/convert_s8d24_to_abgr8.frag index 31db7d426..6a457981d 100644 --- a/src/video_core/host_shaders/convert_s8d24_to_abgr8.frag +++ b/src/video_core/host_shaders/convert_s8d24_to_abgr8.frag @@ -3,16 +3,16 @@ #version 450 +precision mediump int; +precision highp float; + layout(binding = 0) uniform sampler2D depth_tex; -layout(binding = 1) uniform isampler2D stencil_tex; +layout(binding = 1) uniform usampler2D stencil_tex; layout(location = 0) out vec4 color; void main() { ivec2 coord = ivec2(gl_FragCoord.xy); - uint depth = uint(textureLod(depth_tex, coord, 0).r * (exp2(24.0) - 1.0f)); - uint stencil = uint(textureLod(stencil_tex, coord, 0).r); - highp uint depth_val = uint(textureLod(depth_tex, coord, 0).r * (exp2(32.0) - 1.0)); lowp uint stencil_val = textureLod(stencil_tex, coord, 0).r; diff --git a/src/video_core/renderer_vulkan/vk_query_cache.cpp b/src/video_core/renderer_vulkan/vk_query_cache.cpp index 2edaafa7e..66c03bf17 100644 --- a/src/video_core/renderer_vulkan/vk_query_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_query_cache.cpp @@ -1436,6 +1436,7 @@ void QueryCacheRuntime::Barriers(bool is_prebarrier) { .srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT, .dstAccessMask = VK_ACCESS_MEMORY_READ_BIT | VK_ACCESS_MEMORY_WRITE_BIT, }; + impl->scheduler.RequestOutsideRenderPassOperationContext(); if (is_prebarrier) { impl->scheduler.Record([](vk::CommandBuffer cmdbuf) { cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index 83f2b6045..61d03daae 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -975,6 +975,19 @@ void RasterizerVulkan::UpdateScissorsState(Tegra::Engines::Maxwell3D::Regs& regs if (!state_tracker.TouchScissors()) { return; } + if (!regs.viewport_scale_offset_enabled) { + const auto x = static_cast<float>(regs.surface_clip.x); + const auto y = static_cast<float>(regs.surface_clip.y); + const auto width = static_cast<float>(regs.surface_clip.width); + const auto height = static_cast<float>(regs.surface_clip.height); + VkRect2D scissor; + scissor.offset.x = static_cast<u32>(x); + scissor.offset.y = static_cast<u32>(y); + scissor.extent.width = static_cast<u32>(width != 0.0f ? width : 1.0f); + scissor.extent.height = static_cast<u32>(height != 0.0f ? height : 1.0f); + scheduler.Record([scissor](vk::CommandBuffer cmdbuf) { cmdbuf.SetScissor(0, scissor); }); + return; + } u32 up_scale = 1; u32 down_shift = 0; if (texture_cache.IsRescaling()) { diff --git a/src/video_core/renderer_vulkan/vk_render_pass_cache.cpp b/src/video_core/renderer_vulkan/vk_render_pass_cache.cpp index ae9f1de64..7746a88d3 100644 --- a/src/video_core/renderer_vulkan/vk_render_pass_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_render_pass_cache.cpp @@ -19,7 +19,7 @@ VkAttachmentDescription AttachmentDescription(const Device& device, PixelFormat VkSampleCountFlagBits samples) { using MaxwellToVK::SurfaceFormat; return { - .flags = VK_ATTACHMENT_DESCRIPTION_MAY_ALIAS_BIT, + .flags = {}, .format = SurfaceFormat(device, FormatType::Optimal, true, format).format, .samples = samples, .loadOp = VK_ATTACHMENT_LOAD_OP_LOAD, diff --git a/src/video_core/texture_cache/formatter.cpp b/src/video_core/texture_cache/formatter.cpp index 6279d8e9e..2b7e0df72 100644 --- a/src/video_core/texture_cache/formatter.cpp +++ b/src/video_core/texture_cache/formatter.cpp @@ -10,19 +10,23 @@ #include "video_core/texture_cache/image_info.h" #include "video_core/texture_cache/image_view_base.h" #include "video_core/texture_cache/render_targets.h" +#include "video_core/texture_cache/samples_helper.h" namespace VideoCommon { std::string Name(const ImageBase& image) { const GPUVAddr gpu_addr = image.gpu_addr; const ImageInfo& info = image.info; - const u32 width = info.size.width; - const u32 height = info.size.height; + u32 width = info.size.width; + u32 height = info.size.height; const u32 depth = info.size.depth; const u32 num_layers = image.info.resources.layers; const u32 num_levels = image.info.resources.levels; std::string resource; if (image.info.num_samples > 1) { + const auto [samples_x, samples_y] = VideoCommon::SamplesLog2(image.info.num_samples); + width >>= samples_x; + height >>= samples_y; resource += fmt::format(":{}xMSAA", image.info.num_samples); } if (num_layers > 1) { diff --git a/src/video_core/texture_cache/samples_helper.h b/src/video_core/texture_cache/samples_helper.h index 203ac1b11..2ee2f8312 100644 --- a/src/video_core/texture_cache/samples_helper.h +++ b/src/video_core/texture_cache/samples_helper.h @@ -24,7 +24,7 @@ namespace VideoCommon { return {2, 2}; } ASSERT_MSG(false, "Invalid number of samples={}", num_samples); - return {1, 1}; + return {0, 0}; } [[nodiscard]] inline int NumSamples(Tegra::Texture::MsaaMode msaa_mode) { diff --git a/src/video_core/texture_cache/util.cpp b/src/video_core/texture_cache/util.cpp index 8151cabf0..15596c925 100644 --- a/src/video_core/texture_cache/util.cpp +++ b/src/video_core/texture_cache/util.cpp @@ -167,6 +167,13 @@ template <u32 GOB_EXTENT> } [[nodiscard]] constexpr Extent3D TileShift(const LevelInfo& info, u32 level) { + if (level == 0 && info.num_levels == 1) { + return Extent3D{ + .width = info.block.width, + .height = info.block.height, + .depth = info.block.depth, + }; + } const Extent3D blocks = NumLevelBlocks(info, level); return Extent3D{ .width = AdjustTileSize(info.block.width, GOB_SIZE_X, blocks.width), @@ -1293,9 +1300,9 @@ u32 MapSizeBytes(const ImageBase& image) { static_assert(CalculateLevelSize(LevelInfo{{1920, 1080, 1}, {0, 2, 0}, {1, 1}, 2, 0, 1}, 0) == 0x7f8000); -static_assert(CalculateLevelSize(LevelInfo{{32, 32, 1}, {0, 0, 4}, {1, 1}, 4, 0, 1}, 0) == 0x4000); +static_assert(CalculateLevelSize(LevelInfo{{32, 32, 1}, {0, 0, 4}, {1, 1}, 4, 0, 1}, 0) == 0x40000); -static_assert(CalculateLevelSize(LevelInfo{{128, 8, 1}, {0, 4, 0}, {1, 1}, 4, 0, 1}, 0) == 0x4000); +static_assert(CalculateLevelSize(LevelInfo{{128, 8, 1}, {0, 4, 0}, {1, 1}, 4, 0, 1}, 0) == 0x40000); static_assert(CalculateLevelOffset(PixelFormat::R8_SINT, {1920, 1080, 1}, {0, 2, 0}, 0, 7) == 0x2afc00); |