diff options
Diffstat (limited to 'src/video_core/renderer_opengl')
7 files changed, 39 insertions, 18 deletions
diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.cpp b/src/video_core/renderer_opengl/gl_buffer_cache.cpp index d1284e62f..ed188b435 100644 --- a/src/video_core/renderer_opengl/gl_buffer_cache.cpp +++ b/src/video_core/renderer_opengl/gl_buffer_cache.cpp @@ -182,13 +182,14 @@ void BufferCacheRuntime::CopyBuffer(GLuint dst_buffer, Buffer& src_buffer, } void BufferCacheRuntime::CopyBuffer(Buffer& dst_buffer, GLuint src_buffer, - std::span<const VideoCommon::BufferCopy> copies, bool barrier) { + std::span<const VideoCommon::BufferCopy> copies, bool barrier, + bool) { CopyBuffer(dst_buffer.Handle(), src_buffer, copies, barrier); } void BufferCacheRuntime::CopyBuffer(Buffer& dst_buffer, Buffer& src_buffer, - std::span<const VideoCommon::BufferCopy> copies) { - CopyBuffer(dst_buffer.Handle(), src_buffer.Handle(), copies); + std::span<const VideoCommon::BufferCopy> copies, bool) { + CopyBuffer(dst_buffer.Handle(), src_buffer.Handle(), copies, true); } void BufferCacheRuntime::PreCopyBarrier() { diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.h b/src/video_core/renderer_opengl/gl_buffer_cache.h index 1b87954d7..1e8708f59 100644 --- a/src/video_core/renderer_opengl/gl_buffer_cache.h +++ b/src/video_core/renderer_opengl/gl_buffer_cache.h @@ -30,6 +30,8 @@ public: void MakeResident(GLenum access) noexcept; + void MarkUsage(u64 offset, u64 size) {} + [[nodiscard]] GLuint View(u32 offset, u32 size, VideoCore::Surface::PixelFormat format); [[nodiscard]] GLuint64EXT HostGpuAddr() const noexcept { @@ -68,22 +70,29 @@ public: void FreeDeferredStagingBuffer(StagingBufferMap& buffer); + bool CanReorderUpload(const Buffer&, std::span<const VideoCommon::BufferCopy>) { + return false; + } + void CopyBuffer(GLuint dst_buffer, GLuint src_buffer, - std::span<const VideoCommon::BufferCopy> copies, bool barrier = true); + std::span<const VideoCommon::BufferCopy> copies, bool barrier); void CopyBuffer(GLuint dst_buffer, Buffer& src_buffer, - std::span<const VideoCommon::BufferCopy> copies, bool barrier = true); + std::span<const VideoCommon::BufferCopy> copies, bool barrier); void CopyBuffer(Buffer& dst_buffer, GLuint src_buffer, - std::span<const VideoCommon::BufferCopy> copies, bool barrier = true); + std::span<const VideoCommon::BufferCopy> copies, bool barrier, + bool can_reorder_upload = false); void CopyBuffer(Buffer& dst_buffer, Buffer& src_buffer, - std::span<const VideoCommon::BufferCopy> copies); + std::span<const VideoCommon::BufferCopy> copies, bool); void PreCopyBarrier(); void PostCopyBarrier(); void Finish(); + void TickFrame(VideoCommon::SlotVector<Buffer>&) noexcept {} + void ClearBuffer(Buffer& dest_buffer, u32 offset, size_t size, u32 value); void BindIndexBuffer(Buffer& buffer, u32 offset, u32 size); @@ -184,6 +193,10 @@ public: return device.CanReportMemoryUsage(); } + u32 GetStorageBufferAlignment() const { + return static_cast<u32>(device.GetShaderStorageBufferAlignment()); + } + private: static constexpr std::array PABO_LUT{ GL_VERTEX_PROGRAM_PARAMETER_BUFFER_NV, GL_TESS_CONTROL_PROGRAM_PARAMETER_BUFFER_NV, diff --git a/src/video_core/renderer_opengl/gl_device.cpp b/src/video_core/renderer_opengl/gl_device.cpp index 94258ccd0..993438a27 100644 --- a/src/video_core/renderer_opengl/gl_device.cpp +++ b/src/video_core/renderer_opengl/gl_device.cpp @@ -265,33 +265,33 @@ std::string Device::GetVendorName() const { if (vendor_name == "Intel") { // For Mesa, `Intel` is an overloaded vendor string that could mean crocus or iris. // Simply return `INTEL` for those as well as the Windows driver. - return "INTEL"; + return "Intel"; } if (vendor_name == "Intel Open Source Technology Center") { - return "I965"; + return "i965"; } if (vendor_name == "Mesa Project") { - return "I915"; + return "i915"; } if (vendor_name == "Mesa/X.org") { // This vendor string is overloaded between llvmpipe, softpipe, and virgl, so just return // MESA instead of one of those driver names. - return "MESA"; + return "Mesa"; } if (vendor_name == "AMD") { - return "RADEONSI"; + return "RadeonSI"; } if (vendor_name == "nouveau") { - return "NOUVEAU"; + return "Nouveau"; } if (vendor_name == "X.Org") { return "R600"; } if (vendor_name == "Collabora Ltd") { - return "ZINK"; + return "Zink"; } if (vendor_name == "Intel Corporation") { - return "OPENSWR"; + return "OpenSWR"; } if (vendor_name == "Microsoft Corporation") { return "D3D12"; @@ -300,7 +300,7 @@ std::string Device::GetVendorName() const { // Mesa's tegra driver reports `NVIDIA`. Only present in this list because the default // strategy would have returned `NVIDIA` here for this driver, the same result as the // proprietary driver. - return "TEGRA"; + return "Tegra"; } return vendor_name; } diff --git a/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp b/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp index 44a771d65..af0a453ee 100644 --- a/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp +++ b/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp @@ -559,7 +559,9 @@ void GraphicsPipeline::ConfigureImpl(bool is_indexed) { } void GraphicsPipeline::ConfigureTransformFeedbackImpl() const { - glTransformFeedbackAttribsNV(num_xfb_attribs, xfb_attribs.data(), GL_SEPARATE_ATTRIBS); + const GLenum buffer_mode = + num_xfb_buffers_active == 1 ? GL_INTERLEAVED_ATTRIBS : GL_SEPARATE_ATTRIBS; + glTransformFeedbackAttribsNV(num_xfb_attribs, xfb_attribs.data(), buffer_mode); } void GraphicsPipeline::GenerateTransformFeedbackState() { @@ -567,12 +569,14 @@ void GraphicsPipeline::GenerateTransformFeedbackState() { // when this is required. GLint* cursor{xfb_attribs.data()}; + num_xfb_buffers_active = 0; for (size_t feedback = 0; feedback < Maxwell::NumTransformFeedbackBuffers; ++feedback) { const auto& layout = key.xfb_state.layouts[feedback]; UNIMPLEMENTED_IF_MSG(layout.stride != layout.varying_count * 4, "Stride padding"); if (layout.varying_count == 0) { continue; } + num_xfb_buffers_active++; const auto& locations = key.xfb_state.varyings[feedback]; std::optional<u32> current_index; diff --git a/src/video_core/renderer_opengl/gl_graphics_pipeline.h b/src/video_core/renderer_opengl/gl_graphics_pipeline.h index 74fc9cc3d..2f70c1ae9 100644 --- a/src/video_core/renderer_opengl/gl_graphics_pipeline.h +++ b/src/video_core/renderer_opengl/gl_graphics_pipeline.h @@ -154,6 +154,7 @@ private: static constexpr std::size_t XFB_ENTRY_STRIDE = 3; GLsizei num_xfb_attribs{}; + u32 num_xfb_buffers_active{}; std::array<GLint, 128 * XFB_ENTRY_STRIDE * Maxwell::NumTransformFeedbackBuffers> xfb_attribs{}; std::mutex built_mutex; diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 27e2de1bf..9995b6dd4 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -555,7 +555,7 @@ void RasterizerOpenGL::OnCacheInvalidation(VAddr addr, u64 size) { } { std::scoped_lock lock{buffer_cache.mutex}; - buffer_cache.CachedWriteMemory(addr, size); + buffer_cache.WriteMemory(addr, size); } shader_cache.InvalidateRegion(addr, size); } diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp index 2888e0238..26f2d0ea7 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp @@ -232,6 +232,7 @@ ShaderCache::ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindo .has_gl_bool_ref_bug = device.HasBoolRefBug(), .ignore_nan_fp_comparisons = true, .gl_max_compute_smem_size = device.GetMaxComputeSharedMemorySize(), + .min_ssbo_alignment = device.GetShaderStorageBufferAlignment(), }, host_info{ .support_float64 = true, @@ -240,6 +241,7 @@ ShaderCache::ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindo .needs_demote_reorder = device.IsAmd(), .support_snorm_render_buffer = false, .support_viewport_index_layer = device.HasVertexViewportLayer(), + .min_ssbo_alignment = static_cast<u32>(device.GetShaderStorageBufferAlignment()), .support_geometry_shader_passthrough = device.HasGeometryShaderPassthrough(), .support_conditional_barrier = device.SupportsConditionalBarriers(), } { |