diff options
Diffstat (limited to '')
26 files changed, 768 insertions, 628 deletions
diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp index 733b454de..f85ed8e5b 100644 --- a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp +++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp @@ -34,14 +34,15 @@ constexpr std::array POLYGON_OFFSET_ENABLE_LUT = { }; void RefreshXfbState(VideoCommon::TransformFeedbackState& state, const Maxwell& regs) { - std::ranges::transform(regs.tfb_layouts, state.layouts.begin(), [](const auto& layout) { - return VideoCommon::TransformFeedbackState::Layout{ - .stream = layout.stream, - .varying_count = layout.varying_count, - .stride = layout.stride, - }; - }); - state.varyings = regs.tfb_varying_locs; + std::ranges::transform(regs.transform_feedback.controls, state.layouts.begin(), + [](const auto& layout) { + return VideoCommon::TransformFeedbackState::Layout{ + .stream = layout.stream, + .varying_count = layout.varying_count, + .stride = layout.stride, + }; + }); + state.varyings = regs.stream_out_layout; } } // Anonymous namespace @@ -58,33 +59,38 @@ void FixedPipelineState::Refresh(Tegra::Engines::Maxwell3D& maxwell3d, raw1 = 0; extended_dynamic_state.Assign(has_extended_dynamic_state ? 1 : 0); dynamic_vertex_input.Assign(has_dynamic_vertex_input ? 1 : 0); - xfb_enabled.Assign(regs.tfb_enabled != 0); + xfb_enabled.Assign(regs.transform_feedback_enabled != 0); primitive_restart_enable.Assign(regs.primitive_restart.enabled != 0 ? 1 : 0); depth_bias_enable.Assign(enabled_lut[POLYGON_OFFSET_ENABLE_LUT[topology_index]] != 0 ? 1 : 0); - depth_clamp_disabled.Assign(regs.view_volume_clip_control.depth_clamp_disabled.Value()); + depth_clamp_disabled.Assign(regs.viewport_clip_control.geometry_clip == + Maxwell::ViewportClipControl::GeometryClip::Passthrough || + regs.viewport_clip_control.geometry_clip == + Maxwell::ViewportClipControl::GeometryClip::FrustumXYZ || + regs.viewport_clip_control.geometry_clip == + Maxwell::ViewportClipControl::GeometryClip::FrustumZ); ndc_minus_one_to_one.Assign(regs.depth_mode == Maxwell::DepthMode::MinusOneToOne ? 1 : 0); polygon_mode.Assign(PackPolygonMode(regs.polygon_mode_front)); patch_control_points_minus_one.Assign(regs.patch_vertices - 1); - tessellation_primitive.Assign(static_cast<u32>(regs.tess_mode.prim.Value())); - tessellation_spacing.Assign(static_cast<u32>(regs.tess_mode.spacing.Value())); - tessellation_clockwise.Assign(regs.tess_mode.cw.Value()); + tessellation_primitive.Assign(static_cast<u32>(regs.tessellation.params.domain_type.Value())); + tessellation_spacing.Assign(static_cast<u32>(regs.tessellation.params.spacing.Value())); + tessellation_clockwise.Assign(regs.tessellation.params.output_primitives.Value() == + Maxwell::Tessellation::OutputPrimitives::Triangles_CW); logic_op_enable.Assign(regs.logic_op.enable != 0 ? 1 : 0); - logic_op.Assign(PackLogicOp(regs.logic_op.operation)); + logic_op.Assign(PackLogicOp(regs.logic_op.op)); topology.Assign(regs.draw.topology); - msaa_mode.Assign(regs.multisample_mode); + msaa_mode.Assign(regs.anti_alias_samples_mode); raw2 = 0; rasterize_enable.Assign(regs.rasterize_enable != 0 ? 1 : 0); const auto test_func = - regs.alpha_test_enabled != 0 ? regs.alpha_test_func : Maxwell::ComparisonOp::Always; + regs.alpha_test_enabled != 0 ? regs.alpha_test_func : Maxwell::ComparisonOp::Always_GL; alpha_test_func.Assign(PackComparisonOp(test_func)); - early_z.Assign(regs.force_early_fragment_tests != 0 ? 1 : 0); + early_z.Assign(regs.mandated_early_z != 0 ? 1 : 0); depth_enabled.Assign(regs.zeta_enable != 0 ? 1 : 0); depth_format.Assign(static_cast<u32>(regs.zeta.format)); - y_negate.Assign(regs.screen_y_control.y_negate != 0 ? 1 : 0); - provoking_vertex_last.Assign(regs.provoking_vertex_last != 0 ? 1 : 0); - conservative_raster_enable.Assign(regs.conservative_raster_enable != 0 ? 1 : 0); - smooth_lines.Assign(regs.line_smooth_enable != 0 ? 1 : 0); + y_negate.Assign(regs.window_origin.mode != Maxwell::WindowOrigin::Mode::UpperLeft ? 1 : 0); + provoking_vertex_last.Assign(regs.provoking_vertex == Maxwell::ProvokingVertex::Last ? 1 : 0); + smooth_lines.Assign(regs.line_anti_alias_enable != 0 ? 1 : 0); for (size_t i = 0; i < regs.rt.size(); ++i) { color_formats[i] = static_cast<u8>(regs.rt[i].format); @@ -116,8 +122,8 @@ void FixedPipelineState::Refresh(Tegra::Engines::Maxwell3D& maxwell3d, maxwell3d.dirty.flags[Dirty::VertexInput] = false; enabled_divisors = 0; for (size_t index = 0; index < Maxwell::NumVertexArrays; ++index) { - const bool is_enabled = regs.instanced_arrays.IsInstancingEnabled(index); - binding_divisors[index] = is_enabled ? regs.vertex_array[index].divisor : 0; + const bool is_enabled = regs.vertex_stream_instances.IsInstancingEnabled(index); + binding_divisors[index] = is_enabled ? regs.vertex_streams[index].frequency : 0; enabled_divisors |= (is_enabled ? u64{1} : 0) << index; } for (size_t index = 0; index < Maxwell::NumVertexAttributes; ++index) { @@ -164,17 +170,17 @@ void FixedPipelineState::BlendingAttachment::Refresh(const Maxwell& regs, size_t // TODO: C++20 Use templated lambda to deduplicate code - if (!regs.independent_blend_enable) { - const auto& src = regs.blend; - if (!src.enable[index]) { + if (!regs.blend_per_target_enabled) { + if (!regs.blend.enable[index]) { return; } - equation_rgb.Assign(PackBlendEquation(src.equation_rgb)); - equation_a.Assign(PackBlendEquation(src.equation_a)); - factor_source_rgb.Assign(PackBlendFactor(src.factor_source_rgb)); - factor_dest_rgb.Assign(PackBlendFactor(src.factor_dest_rgb)); - factor_source_a.Assign(PackBlendFactor(src.factor_source_a)); - factor_dest_a.Assign(PackBlendFactor(src.factor_dest_a)); + const auto& src = regs.blend; + equation_rgb.Assign(PackBlendEquation(src.color_op)); + equation_a.Assign(PackBlendEquation(src.alpha_op)); + factor_source_rgb.Assign(PackBlendFactor(src.color_source)); + factor_dest_rgb.Assign(PackBlendFactor(src.color_dest)); + factor_source_a.Assign(PackBlendFactor(src.alpha_source)); + factor_dest_a.Assign(PackBlendFactor(src.alpha_dest)); enable.Assign(1); return; } @@ -182,34 +188,34 @@ void FixedPipelineState::BlendingAttachment::Refresh(const Maxwell& regs, size_t if (!regs.blend.enable[index]) { return; } - const auto& src = regs.independent_blend[index]; - equation_rgb.Assign(PackBlendEquation(src.equation_rgb)); - equation_a.Assign(PackBlendEquation(src.equation_a)); - factor_source_rgb.Assign(PackBlendFactor(src.factor_source_rgb)); - factor_dest_rgb.Assign(PackBlendFactor(src.factor_dest_rgb)); - factor_source_a.Assign(PackBlendFactor(src.factor_source_a)); - factor_dest_a.Assign(PackBlendFactor(src.factor_dest_a)); + const auto& src = regs.blend_per_target[index]; + equation_rgb.Assign(PackBlendEquation(src.color_op)); + equation_a.Assign(PackBlendEquation(src.alpha_op)); + factor_source_rgb.Assign(PackBlendFactor(src.color_source)); + factor_dest_rgb.Assign(PackBlendFactor(src.color_dest)); + factor_source_a.Assign(PackBlendFactor(src.alpha_source)); + factor_dest_a.Assign(PackBlendFactor(src.alpha_dest)); enable.Assign(1); } void FixedPipelineState::DynamicState::Refresh(const Maxwell& regs) { - u32 packed_front_face = PackFrontFace(regs.front_face); - if (regs.screen_y_control.triangle_rast_flip != 0) { + u32 packed_front_face = PackFrontFace(regs.gl_front_face); + if (regs.window_origin.flip_y != 0) { // Flip front face packed_front_face = 1 - packed_front_face; } raw1 = 0; raw2 = 0; - front.action_stencil_fail.Assign(PackStencilOp(regs.stencil_front_op_fail)); - front.action_depth_fail.Assign(PackStencilOp(regs.stencil_front_op_zfail)); - front.action_depth_pass.Assign(PackStencilOp(regs.stencil_front_op_zpass)); - front.test_func.Assign(PackComparisonOp(regs.stencil_front_func_func)); + front.action_stencil_fail.Assign(PackStencilOp(regs.stencil_front_op.fail)); + front.action_depth_fail.Assign(PackStencilOp(regs.stencil_front_op.zfail)); + front.action_depth_pass.Assign(PackStencilOp(regs.stencil_front_op.zpass)); + front.test_func.Assign(PackComparisonOp(regs.stencil_front_op.func)); if (regs.stencil_two_side_enable) { - back.action_stencil_fail.Assign(PackStencilOp(regs.stencil_back_op_fail)); - back.action_depth_fail.Assign(PackStencilOp(regs.stencil_back_op_zfail)); - back.action_depth_pass.Assign(PackStencilOp(regs.stencil_back_op_zpass)); - back.test_func.Assign(PackComparisonOp(regs.stencil_back_func_func)); + back.action_stencil_fail.Assign(PackStencilOp(regs.stencil_back_op.fail)); + back.action_depth_fail.Assign(PackStencilOp(regs.stencil_back_op.zfail)); + back.action_depth_pass.Assign(PackStencilOp(regs.stencil_back_op.zpass)); + back.test_func.Assign(PackComparisonOp(regs.stencil_back_op.func)); } else { back.action_stencil_fail.Assign(front.action_stencil_fail); back.action_depth_fail.Assign(front.action_depth_fail); @@ -222,9 +228,9 @@ void FixedPipelineState::DynamicState::Refresh(const Maxwell& regs) { depth_test_enable.Assign(regs.depth_test_enable); front_face.Assign(packed_front_face); depth_test_func.Assign(PackComparisonOp(regs.depth_test_func)); - cull_face.Assign(PackCullFace(regs.cull_face)); - cull_enable.Assign(regs.cull_test_enabled != 0 ? 1 : 0); - std::ranges::transform(regs.vertex_array, vertex_strides.begin(), [](const auto& array) { + cull_face.Assign(PackCullFace(regs.gl_cull_face)); + cull_enable.Assign(regs.gl_cull_test_enabled != 0 ? 1 : 0); + std::ranges::transform(regs.vertex_streams, vertex_strides.begin(), [](const auto& array) { return static_cast<u16>(array.stride.Value()); }); } @@ -251,41 +257,42 @@ Maxwell::ComparisonOp FixedPipelineState::UnpackComparisonOp(u32 packed) noexcep return static_cast<Maxwell::ComparisonOp>(packed + 1); } -u32 FixedPipelineState::PackStencilOp(Maxwell::StencilOp op) noexcept { +u32 FixedPipelineState::PackStencilOp(Maxwell::StencilOp::Op op) noexcept { switch (op) { - case Maxwell::StencilOp::Keep: - case Maxwell::StencilOp::KeepOGL: + case Maxwell::StencilOp::Op::Keep_D3D: + case Maxwell::StencilOp::Op::Keep_GL: return 0; - case Maxwell::StencilOp::Zero: - case Maxwell::StencilOp::ZeroOGL: + case Maxwell::StencilOp::Op::Zero_D3D: + case Maxwell::StencilOp::Op::Zero_GL: return 1; - case Maxwell::StencilOp::Replace: - case Maxwell::StencilOp::ReplaceOGL: + case Maxwell::StencilOp::Op::Replace_D3D: + case Maxwell::StencilOp::Op::Replace_GL: return 2; - case Maxwell::StencilOp::Incr: - case Maxwell::StencilOp::IncrOGL: + case Maxwell::StencilOp::Op::IncrSaturate_D3D: + case Maxwell::StencilOp::Op::IncrSaturate_GL: return 3; - case Maxwell::StencilOp::Decr: - case Maxwell::StencilOp::DecrOGL: + case Maxwell::StencilOp::Op::DecrSaturate_D3D: + case Maxwell::StencilOp::Op::DecrSaturate_GL: return 4; - case Maxwell::StencilOp::Invert: - case Maxwell::StencilOp::InvertOGL: + case Maxwell::StencilOp::Op::Invert_D3D: + case Maxwell::StencilOp::Op::Invert_GL: return 5; - case Maxwell::StencilOp::IncrWrap: - case Maxwell::StencilOp::IncrWrapOGL: + case Maxwell::StencilOp::Op::Incr_D3D: + case Maxwell::StencilOp::Op::Incr_GL: return 6; - case Maxwell::StencilOp::DecrWrap: - case Maxwell::StencilOp::DecrWrapOGL: + case Maxwell::StencilOp::Op::Decr_D3D: + case Maxwell::StencilOp::Op::Decr_GL: return 7; } return 0; } -Maxwell::StencilOp FixedPipelineState::UnpackStencilOp(u32 packed) noexcept { - static constexpr std::array LUT = {Maxwell::StencilOp::Keep, Maxwell::StencilOp::Zero, - Maxwell::StencilOp::Replace, Maxwell::StencilOp::Incr, - Maxwell::StencilOp::Decr, Maxwell::StencilOp::Invert, - Maxwell::StencilOp::IncrWrap, Maxwell::StencilOp::DecrWrap}; +Maxwell::StencilOp::Op FixedPipelineState::UnpackStencilOp(u32 packed) noexcept { + static constexpr std::array LUT = { + Maxwell::StencilOp::Op::Keep_D3D, Maxwell::StencilOp::Op::Zero_D3D, + Maxwell::StencilOp::Op::Replace_D3D, Maxwell::StencilOp::Op::IncrSaturate_D3D, + Maxwell::StencilOp::Op::DecrSaturate_D3D, Maxwell::StencilOp::Op::Invert_D3D, + Maxwell::StencilOp::Op::Incr_D3D, Maxwell::StencilOp::Op::Decr_D3D}; return LUT[packed]; } @@ -318,30 +325,30 @@ Maxwell::PolygonMode FixedPipelineState::UnpackPolygonMode(u32 packed) noexcept return static_cast<Maxwell::PolygonMode>(packed + 0x1B00); } -u32 FixedPipelineState::PackLogicOp(Maxwell::LogicOperation op) noexcept { +u32 FixedPipelineState::PackLogicOp(Maxwell::LogicOp::Op op) noexcept { return static_cast<u32>(op) - 0x1500; } -Maxwell::LogicOperation FixedPipelineState::UnpackLogicOp(u32 packed) noexcept { - return static_cast<Maxwell::LogicOperation>(packed + 0x1500); +Maxwell::LogicOp::Op FixedPipelineState::UnpackLogicOp(u32 packed) noexcept { + return static_cast<Maxwell::LogicOp::Op>(packed + 0x1500); } u32 FixedPipelineState::PackBlendEquation(Maxwell::Blend::Equation equation) noexcept { switch (equation) { - case Maxwell::Blend::Equation::Add: - case Maxwell::Blend::Equation::AddGL: + case Maxwell::Blend::Equation::Add_D3D: + case Maxwell::Blend::Equation::Add_GL: return 0; - case Maxwell::Blend::Equation::Subtract: - case Maxwell::Blend::Equation::SubtractGL: + case Maxwell::Blend::Equation::Subtract_D3D: + case Maxwell::Blend::Equation::Subtract_GL: return 1; - case Maxwell::Blend::Equation::ReverseSubtract: - case Maxwell::Blend::Equation::ReverseSubtractGL: + case Maxwell::Blend::Equation::ReverseSubtract_D3D: + case Maxwell::Blend::Equation::ReverseSubtract_GL: return 2; - case Maxwell::Blend::Equation::Min: - case Maxwell::Blend::Equation::MinGL: + case Maxwell::Blend::Equation::Min_D3D: + case Maxwell::Blend::Equation::Min_GL: return 3; - case Maxwell::Blend::Equation::Max: - case Maxwell::Blend::Equation::MaxGL: + case Maxwell::Blend::Equation::Max_D3D: + case Maxwell::Blend::Equation::Max_GL: return 4; } return 0; @@ -349,97 +356,99 @@ u32 FixedPipelineState::PackBlendEquation(Maxwell::Blend::Equation equation) noe Maxwell::Blend::Equation FixedPipelineState::UnpackBlendEquation(u32 packed) noexcept { static constexpr std::array LUT = { - Maxwell::Blend::Equation::Add, Maxwell::Blend::Equation::Subtract, - Maxwell::Blend::Equation::ReverseSubtract, Maxwell::Blend::Equation::Min, - Maxwell::Blend::Equation::Max}; + Maxwell::Blend::Equation::Add_D3D, Maxwell::Blend::Equation::Subtract_D3D, + Maxwell::Blend::Equation::ReverseSubtract_D3D, Maxwell::Blend::Equation::Min_D3D, + Maxwell::Blend::Equation::Max_D3D}; return LUT[packed]; } u32 FixedPipelineState::PackBlendFactor(Maxwell::Blend::Factor factor) noexcept { switch (factor) { - case Maxwell::Blend::Factor::Zero: - case Maxwell::Blend::Factor::ZeroGL: + case Maxwell::Blend::Factor::Zero_D3D: + case Maxwell::Blend::Factor::Zero_GL: return 0; - case Maxwell::Blend::Factor::One: - case Maxwell::Blend::Factor::OneGL: + case Maxwell::Blend::Factor::One_D3D: + case Maxwell::Blend::Factor::One_GL: return 1; - case Maxwell::Blend::Factor::SourceColor: - case Maxwell::Blend::Factor::SourceColorGL: + case Maxwell::Blend::Factor::SourceColor_D3D: + case Maxwell::Blend::Factor::SourceColor_GL: return 2; - case Maxwell::Blend::Factor::OneMinusSourceColor: - case Maxwell::Blend::Factor::OneMinusSourceColorGL: + case Maxwell::Blend::Factor::OneMinusSourceColor_D3D: + case Maxwell::Blend::Factor::OneMinusSourceColor_GL: return 3; - case Maxwell::Blend::Factor::SourceAlpha: - case Maxwell::Blend::Factor::SourceAlphaGL: + case Maxwell::Blend::Factor::SourceAlpha_D3D: + case Maxwell::Blend::Factor::SourceAlpha_GL: return 4; - case Maxwell::Blend::Factor::OneMinusSourceAlpha: - case Maxwell::Blend::Factor::OneMinusSourceAlphaGL: + case Maxwell::Blend::Factor::OneMinusSourceAlpha_D3D: + case Maxwell::Blend::Factor::OneMinusSourceAlpha_GL: return 5; - case Maxwell::Blend::Factor::DestAlpha: - case Maxwell::Blend::Factor::DestAlphaGL: + case Maxwell::Blend::Factor::DestAlpha_D3D: + case Maxwell::Blend::Factor::DestAlpha_GL: return 6; - case Maxwell::Blend::Factor::OneMinusDestAlpha: - case Maxwell::Blend::Factor::OneMinusDestAlphaGL: + case Maxwell::Blend::Factor::OneMinusDestAlpha_D3D: + case Maxwell::Blend::Factor::OneMinusDestAlpha_GL: return 7; - case Maxwell::Blend::Factor::DestColor: - case Maxwell::Blend::Factor::DestColorGL: + case Maxwell::Blend::Factor::DestColor_D3D: + case Maxwell::Blend::Factor::DestColor_GL: return 8; - case Maxwell::Blend::Factor::OneMinusDestColor: - case Maxwell::Blend::Factor::OneMinusDestColorGL: + case Maxwell::Blend::Factor::OneMinusDestColor_D3D: + case Maxwell::Blend::Factor::OneMinusDestColor_GL: return 9; - case Maxwell::Blend::Factor::SourceAlphaSaturate: - case Maxwell::Blend::Factor::SourceAlphaSaturateGL: + case Maxwell::Blend::Factor::SourceAlphaSaturate_D3D: + case Maxwell::Blend::Factor::SourceAlphaSaturate_GL: return 10; - case Maxwell::Blend::Factor::Source1Color: - case Maxwell::Blend::Factor::Source1ColorGL: + case Maxwell::Blend::Factor::Source1Color_D3D: + case Maxwell::Blend::Factor::Source1Color_GL: return 11; - case Maxwell::Blend::Factor::OneMinusSource1Color: - case Maxwell::Blend::Factor::OneMinusSource1ColorGL: + case Maxwell::Blend::Factor::OneMinusSource1Color_D3D: + case Maxwell::Blend::Factor::OneMinusSource1Color_GL: return 12; - case Maxwell::Blend::Factor::Source1Alpha: - case Maxwell::Blend::Factor::Source1AlphaGL: + case Maxwell::Blend::Factor::Source1Alpha_D3D: + case Maxwell::Blend::Factor::Source1Alpha_GL: return 13; - case Maxwell::Blend::Factor::OneMinusSource1Alpha: - case Maxwell::Blend::Factor::OneMinusSource1AlphaGL: + case Maxwell::Blend::Factor::OneMinusSource1Alpha_D3D: + case Maxwell::Blend::Factor::OneMinusSource1Alpha_GL: return 14; - case Maxwell::Blend::Factor::ConstantColor: - case Maxwell::Blend::Factor::ConstantColorGL: + case Maxwell::Blend::Factor::BlendFactor_D3D: + case Maxwell::Blend::Factor::ConstantColor_GL: return 15; - case Maxwell::Blend::Factor::OneMinusConstantColor: - case Maxwell::Blend::Factor::OneMinusConstantColorGL: + case Maxwell::Blend::Factor::OneMinusBlendFactor_D3D: + case Maxwell::Blend::Factor::OneMinusConstantColor_GL: return 16; - case Maxwell::Blend::Factor::ConstantAlpha: - case Maxwell::Blend::Factor::ConstantAlphaGL: + case Maxwell::Blend::Factor::BothSourceAlpha_D3D: + case Maxwell::Blend::Factor::ConstantAlpha_GL: return 17; - case Maxwell::Blend::Factor::OneMinusConstantAlpha: - case Maxwell::Blend::Factor::OneMinusConstantAlphaGL: + case Maxwell::Blend::Factor::OneMinusBothSourceAlpha_D3D: + case Maxwell::Blend::Factor::OneMinusConstantAlpha_GL: return 18; } + UNIMPLEMENTED_MSG("Unknown blend factor {}", static_cast<u32>(factor)); return 0; } Maxwell::Blend::Factor FixedPipelineState::UnpackBlendFactor(u32 packed) noexcept { static constexpr std::array LUT = { - Maxwell::Blend::Factor::Zero, - Maxwell::Blend::Factor::One, - Maxwell::Blend::Factor::SourceColor, - Maxwell::Blend::Factor::OneMinusSourceColor, - Maxwell::Blend::Factor::SourceAlpha, - Maxwell::Blend::Factor::OneMinusSourceAlpha, - Maxwell::Blend::Factor::DestAlpha, - Maxwell::Blend::Factor::OneMinusDestAlpha, - Maxwell::Blend::Factor::DestColor, - Maxwell::Blend::Factor::OneMinusDestColor, - Maxwell::Blend::Factor::SourceAlphaSaturate, - Maxwell::Blend::Factor::Source1Color, - Maxwell::Blend::Factor::OneMinusSource1Color, - Maxwell::Blend::Factor::Source1Alpha, - Maxwell::Blend::Factor::OneMinusSource1Alpha, - Maxwell::Blend::Factor::ConstantColor, - Maxwell::Blend::Factor::OneMinusConstantColor, - Maxwell::Blend::Factor::ConstantAlpha, - Maxwell::Blend::Factor::OneMinusConstantAlpha, + Maxwell::Blend::Factor::Zero_D3D, + Maxwell::Blend::Factor::One_D3D, + Maxwell::Blend::Factor::SourceColor_D3D, + Maxwell::Blend::Factor::OneMinusSourceColor_D3D, + Maxwell::Blend::Factor::SourceAlpha_D3D, + Maxwell::Blend::Factor::OneMinusSourceAlpha_D3D, + Maxwell::Blend::Factor::DestAlpha_D3D, + Maxwell::Blend::Factor::OneMinusDestAlpha_D3D, + Maxwell::Blend::Factor::DestColor_D3D, + Maxwell::Blend::Factor::OneMinusDestColor_D3D, + Maxwell::Blend::Factor::SourceAlphaSaturate_D3D, + Maxwell::Blend::Factor::Source1Color_D3D, + Maxwell::Blend::Factor::OneMinusSource1Color_D3D, + Maxwell::Blend::Factor::Source1Alpha_D3D, + Maxwell::Blend::Factor::OneMinusSource1Alpha_D3D, + Maxwell::Blend::Factor::BlendFactor_D3D, + Maxwell::Blend::Factor::OneMinusBlendFactor_D3D, + Maxwell::Blend::Factor::BothSourceAlpha_D3D, + Maxwell::Blend::Factor::OneMinusBothSourceAlpha_D3D, }; + ASSERT(packed < LUT.size()); return LUT[packed]; } diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.h b/src/video_core/renderer_vulkan/fixed_pipeline_state.h index 9d60756e5..43441209c 100644 --- a/src/video_core/renderer_vulkan/fixed_pipeline_state.h +++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.h @@ -21,8 +21,8 @@ struct FixedPipelineState { static u32 PackComparisonOp(Maxwell::ComparisonOp op) noexcept; static Maxwell::ComparisonOp UnpackComparisonOp(u32 packed) noexcept; - static u32 PackStencilOp(Maxwell::StencilOp op) noexcept; - static Maxwell::StencilOp UnpackStencilOp(u32 packed) noexcept; + static u32 PackStencilOp(Maxwell::StencilOp::Op op) noexcept; + static Maxwell::StencilOp::Op UnpackStencilOp(u32 packed) noexcept; static u32 PackCullFace(Maxwell::CullFace cull) noexcept; static Maxwell::CullFace UnpackCullFace(u32 packed) noexcept; @@ -33,8 +33,8 @@ struct FixedPipelineState { static u32 PackPolygonMode(Maxwell::PolygonMode mode) noexcept; static Maxwell::PolygonMode UnpackPolygonMode(u32 packed) noexcept; - static u32 PackLogicOp(Maxwell::LogicOperation op) noexcept; - static Maxwell::LogicOperation UnpackLogicOp(u32 packed) noexcept; + static u32 PackLogicOp(Maxwell::LogicOp::Op op) noexcept; + static Maxwell::LogicOp::Op UnpackLogicOp(u32 packed) noexcept; static u32 PackBlendEquation(Maxwell::Blend::Equation equation) noexcept; static Maxwell::Blend::Equation UnpackBlendEquation(u32 packed) noexcept; @@ -113,15 +113,15 @@ struct FixedPipelineState { BitField<Position + 6, 3, u32> action_depth_pass; BitField<Position + 9, 3, u32> test_func; - Maxwell::StencilOp ActionStencilFail() const noexcept { + Maxwell::StencilOp::Op ActionStencilFail() const noexcept { return UnpackStencilOp(action_stencil_fail); } - Maxwell::StencilOp ActionDepthFail() const noexcept { + Maxwell::StencilOp::Op ActionDepthFail() const noexcept { return UnpackStencilOp(action_depth_fail); } - Maxwell::StencilOp ActionDepthPass() const noexcept { + Maxwell::StencilOp::Op ActionDepthPass() const noexcept { return UnpackStencilOp(action_depth_pass); } @@ -193,7 +193,6 @@ struct FixedPipelineState { BitField<6, 5, u32> depth_format; BitField<11, 1, u32> y_negate; BitField<12, 1, u32> provoking_vertex_last; - BitField<13, 1, u32> conservative_raster_enable; BitField<14, 1, u32> smooth_lines; }; std::array<u8, Maxwell::NumRenderTargets> color_formats; diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp index bdb71dc53..5c156087b 100644 --- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp +++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp @@ -184,7 +184,7 @@ struct FormatTuple { {VK_FORMAT_BC3_SRGB_BLOCK}, // BC3_SRGB {VK_FORMAT_BC7_SRGB_BLOCK}, // BC7_SRGB {VK_FORMAT_R4G4B4A4_UNORM_PACK16, Attachable}, // A4B4G4R4_UNORM - {VK_FORMAT_R4G4_UNORM_PACK8}, // R4G4_UNORM + {VK_FORMAT_R4G4_UNORM_PACK8}, // G4R4_UNORM {VK_FORMAT_ASTC_4x4_SRGB_BLOCK}, // ASTC_2D_4X4_SRGB {VK_FORMAT_ASTC_8x8_SRGB_BLOCK}, // ASTC_2D_8X8_SRGB {VK_FORMAT_ASTC_8x5_SRGB_BLOCK}, // ASTC_2D_8X5_SRGB @@ -196,6 +196,8 @@ struct FormatTuple { {VK_FORMAT_ASTC_6x6_UNORM_BLOCK}, // ASTC_2D_6X6_UNORM {VK_FORMAT_ASTC_6x6_SRGB_BLOCK}, // ASTC_2D_6X6_SRGB {VK_FORMAT_ASTC_10x6_UNORM_BLOCK}, // ASTC_2D_10X6_UNORM + {VK_FORMAT_ASTC_10x5_UNORM_BLOCK}, // ASTC_2D_10X5_UNORM + {VK_FORMAT_ASTC_10x5_SRGB_BLOCK}, // ASTC_2D_10X5_SRGB {VK_FORMAT_ASTC_10x10_UNORM_BLOCK}, // ASTC_2D_10X10_UNORM {VK_FORMAT_ASTC_10x10_SRGB_BLOCK}, // ASTC_2D_10X10_SRGB {VK_FORMAT_ASTC_12x12_UNORM_BLOCK}, // ASTC_2D_12X12_UNORM @@ -321,161 +323,182 @@ VkFormat VertexFormat(const Device& device, Maxwell::VertexAttribute::Type type, Maxwell::VertexAttribute::Size size) { const VkFormat format{([&]() { switch (type) { - case Maxwell::VertexAttribute::Type::UnsignedNorm: + case Maxwell::VertexAttribute::Type::UnusedEnumDoNotUseBecauseItWillGoAway: + ASSERT_MSG(false, "Invalid vertex attribute type!"); + break; + case Maxwell::VertexAttribute::Type::UNorm: switch (size) { - case Maxwell::VertexAttribute::Size::Size_8: + case Maxwell::VertexAttribute::Size::Size_R8: + case Maxwell::VertexAttribute::Size::Size_A8: return VK_FORMAT_R8_UNORM; - case Maxwell::VertexAttribute::Size::Size_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8: + case Maxwell::VertexAttribute::Size::Size_G8_R8: return VK_FORMAT_R8G8_UNORM; - case Maxwell::VertexAttribute::Size::Size_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8: return VK_FORMAT_R8G8B8_UNORM; - case Maxwell::VertexAttribute::Size::Size_8_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8_A8: + case Maxwell::VertexAttribute::Size::Size_X8_B8_G8_R8: return VK_FORMAT_R8G8B8A8_UNORM; - case Maxwell::VertexAttribute::Size::Size_16: + case Maxwell::VertexAttribute::Size::Size_R16: return VK_FORMAT_R16_UNORM; - case Maxwell::VertexAttribute::Size::Size_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16: return VK_FORMAT_R16G16_UNORM; - case Maxwell::VertexAttribute::Size::Size_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16: return VK_FORMAT_R16G16B16_UNORM; - case Maxwell::VertexAttribute::Size::Size_16_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16_A16: return VK_FORMAT_R16G16B16A16_UNORM; - case Maxwell::VertexAttribute::Size::Size_10_10_10_2: + case Maxwell::VertexAttribute::Size::Size_A2_B10_G10_R10: return VK_FORMAT_A2B10G10R10_UNORM_PACK32; default: break; } break; - case Maxwell::VertexAttribute::Type::SignedNorm: + case Maxwell::VertexAttribute::Type::SNorm: switch (size) { - case Maxwell::VertexAttribute::Size::Size_8: + case Maxwell::VertexAttribute::Size::Size_R8: + case Maxwell::VertexAttribute::Size::Size_A8: return VK_FORMAT_R8_SNORM; - case Maxwell::VertexAttribute::Size::Size_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8: + case Maxwell::VertexAttribute::Size::Size_G8_R8: return VK_FORMAT_R8G8_SNORM; - case Maxwell::VertexAttribute::Size::Size_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8: return VK_FORMAT_R8G8B8_SNORM; - case Maxwell::VertexAttribute::Size::Size_8_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8_A8: + case Maxwell::VertexAttribute::Size::Size_X8_B8_G8_R8: return VK_FORMAT_R8G8B8A8_SNORM; - case Maxwell::VertexAttribute::Size::Size_16: + case Maxwell::VertexAttribute::Size::Size_R16: return VK_FORMAT_R16_SNORM; - case Maxwell::VertexAttribute::Size::Size_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16: return VK_FORMAT_R16G16_SNORM; - case Maxwell::VertexAttribute::Size::Size_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16: return VK_FORMAT_R16G16B16_SNORM; - case Maxwell::VertexAttribute::Size::Size_16_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16_A16: return VK_FORMAT_R16G16B16A16_SNORM; - case Maxwell::VertexAttribute::Size::Size_10_10_10_2: + case Maxwell::VertexAttribute::Size::Size_A2_B10_G10_R10: return VK_FORMAT_A2B10G10R10_SNORM_PACK32; default: break; } break; - case Maxwell::VertexAttribute::Type::UnsignedScaled: + case Maxwell::VertexAttribute::Type::UScaled: switch (size) { - case Maxwell::VertexAttribute::Size::Size_8: + case Maxwell::VertexAttribute::Size::Size_R8: + case Maxwell::VertexAttribute::Size::Size_A8: return VK_FORMAT_R8_USCALED; - case Maxwell::VertexAttribute::Size::Size_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8: + case Maxwell::VertexAttribute::Size::Size_G8_R8: return VK_FORMAT_R8G8_USCALED; - case Maxwell::VertexAttribute::Size::Size_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8: return VK_FORMAT_R8G8B8_USCALED; - case Maxwell::VertexAttribute::Size::Size_8_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8_A8: + case Maxwell::VertexAttribute::Size::Size_X8_B8_G8_R8: return VK_FORMAT_R8G8B8A8_USCALED; - case Maxwell::VertexAttribute::Size::Size_16: + case Maxwell::VertexAttribute::Size::Size_R16: return VK_FORMAT_R16_USCALED; - case Maxwell::VertexAttribute::Size::Size_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16: return VK_FORMAT_R16G16_USCALED; - case Maxwell::VertexAttribute::Size::Size_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16: return VK_FORMAT_R16G16B16_USCALED; - case Maxwell::VertexAttribute::Size::Size_16_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16_A16: return VK_FORMAT_R16G16B16A16_USCALED; - case Maxwell::VertexAttribute::Size::Size_10_10_10_2: + case Maxwell::VertexAttribute::Size::Size_A2_B10_G10_R10: return VK_FORMAT_A2B10G10R10_USCALED_PACK32; default: break; } break; - case Maxwell::VertexAttribute::Type::SignedScaled: + case Maxwell::VertexAttribute::Type::SScaled: switch (size) { - case Maxwell::VertexAttribute::Size::Size_8: + case Maxwell::VertexAttribute::Size::Size_R8: + case Maxwell::VertexAttribute::Size::Size_A8: return VK_FORMAT_R8_SSCALED; - case Maxwell::VertexAttribute::Size::Size_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8: + case Maxwell::VertexAttribute::Size::Size_G8_R8: return VK_FORMAT_R8G8_SSCALED; - case Maxwell::VertexAttribute::Size::Size_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8: return VK_FORMAT_R8G8B8_SSCALED; - case Maxwell::VertexAttribute::Size::Size_8_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8_A8: + case Maxwell::VertexAttribute::Size::Size_X8_B8_G8_R8: return VK_FORMAT_R8G8B8A8_SSCALED; - case Maxwell::VertexAttribute::Size::Size_16: + case Maxwell::VertexAttribute::Size::Size_R16: return VK_FORMAT_R16_SSCALED; - case Maxwell::VertexAttribute::Size::Size_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16: return VK_FORMAT_R16G16_SSCALED; - case Maxwell::VertexAttribute::Size::Size_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16: return VK_FORMAT_R16G16B16_SSCALED; - case Maxwell::VertexAttribute::Size::Size_16_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16_A16: return VK_FORMAT_R16G16B16A16_SSCALED; - case Maxwell::VertexAttribute::Size::Size_10_10_10_2: + case Maxwell::VertexAttribute::Size::Size_A2_B10_G10_R10: return VK_FORMAT_A2B10G10R10_SSCALED_PACK32; default: break; } break; - case Maxwell::VertexAttribute::Type::UnsignedInt: + case Maxwell::VertexAttribute::Type::UInt: switch (size) { - case Maxwell::VertexAttribute::Size::Size_8: + case Maxwell::VertexAttribute::Size::Size_R8: + case Maxwell::VertexAttribute::Size::Size_A8: return VK_FORMAT_R8_UINT; - case Maxwell::VertexAttribute::Size::Size_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8: + case Maxwell::VertexAttribute::Size::Size_G8_R8: return VK_FORMAT_R8G8_UINT; - case Maxwell::VertexAttribute::Size::Size_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8: return VK_FORMAT_R8G8B8_UINT; - case Maxwell::VertexAttribute::Size::Size_8_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8_A8: + case Maxwell::VertexAttribute::Size::Size_X8_B8_G8_R8: return VK_FORMAT_R8G8B8A8_UINT; - case Maxwell::VertexAttribute::Size::Size_16: + case Maxwell::VertexAttribute::Size::Size_R16: return VK_FORMAT_R16_UINT; - case Maxwell::VertexAttribute::Size::Size_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16: return VK_FORMAT_R16G16_UINT; - case Maxwell::VertexAttribute::Size::Size_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16: return VK_FORMAT_R16G16B16_UINT; - case Maxwell::VertexAttribute::Size::Size_16_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16_A16: return VK_FORMAT_R16G16B16A16_UINT; - case Maxwell::VertexAttribute::Size::Size_32: + case Maxwell::VertexAttribute::Size::Size_R32: return VK_FORMAT_R32_UINT; - case Maxwell::VertexAttribute::Size::Size_32_32: + case Maxwell::VertexAttribute::Size::Size_R32_G32: return VK_FORMAT_R32G32_UINT; - case Maxwell::VertexAttribute::Size::Size_32_32_32: + case Maxwell::VertexAttribute::Size::Size_R32_G32_B32: return VK_FORMAT_R32G32B32_UINT; - case Maxwell::VertexAttribute::Size::Size_32_32_32_32: + case Maxwell::VertexAttribute::Size::Size_R32_G32_B32_A32: return VK_FORMAT_R32G32B32A32_UINT; - case Maxwell::VertexAttribute::Size::Size_10_10_10_2: + case Maxwell::VertexAttribute::Size::Size_A2_B10_G10_R10: return VK_FORMAT_A2B10G10R10_UINT_PACK32; default: break; } break; - case Maxwell::VertexAttribute::Type::SignedInt: + case Maxwell::VertexAttribute::Type::SInt: switch (size) { - case Maxwell::VertexAttribute::Size::Size_8: + case Maxwell::VertexAttribute::Size::Size_R8: + case Maxwell::VertexAttribute::Size::Size_A8: return VK_FORMAT_R8_SINT; - case Maxwell::VertexAttribute::Size::Size_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8: + case Maxwell::VertexAttribute::Size::Size_G8_R8: return VK_FORMAT_R8G8_SINT; - case Maxwell::VertexAttribute::Size::Size_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8: return VK_FORMAT_R8G8B8_SINT; - case Maxwell::VertexAttribute::Size::Size_8_8_8_8: + case Maxwell::VertexAttribute::Size::Size_R8_G8_B8_A8: + case Maxwell::VertexAttribute::Size::Size_X8_B8_G8_R8: return VK_FORMAT_R8G8B8A8_SINT; - case Maxwell::VertexAttribute::Size::Size_16: + case Maxwell::VertexAttribute::Size::Size_R16: return VK_FORMAT_R16_SINT; - case Maxwell::VertexAttribute::Size::Size_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16: return VK_FORMAT_R16G16_SINT; - case Maxwell::VertexAttribute::Size::Size_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16: return VK_FORMAT_R16G16B16_SINT; - case Maxwell::VertexAttribute::Size::Size_16_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16_A16: return VK_FORMAT_R16G16B16A16_SINT; - case Maxwell::VertexAttribute::Size::Size_32: + case Maxwell::VertexAttribute::Size::Size_R32: return VK_FORMAT_R32_SINT; - case Maxwell::VertexAttribute::Size::Size_32_32: + case Maxwell::VertexAttribute::Size::Size_R32_G32: return VK_FORMAT_R32G32_SINT; - case Maxwell::VertexAttribute::Size::Size_32_32_32: + case Maxwell::VertexAttribute::Size::Size_R32_G32_B32: return VK_FORMAT_R32G32B32_SINT; - case Maxwell::VertexAttribute::Size::Size_32_32_32_32: + case Maxwell::VertexAttribute::Size::Size_R32_G32_B32_A32: return VK_FORMAT_R32G32B32A32_SINT; - case Maxwell::VertexAttribute::Size::Size_10_10_10_2: + case Maxwell::VertexAttribute::Size::Size_A2_B10_G10_R10: return VK_FORMAT_A2B10G10R10_SINT_PACK32; default: break; @@ -483,23 +506,23 @@ VkFormat VertexFormat(const Device& device, Maxwell::VertexAttribute::Type type, break; case Maxwell::VertexAttribute::Type::Float: switch (size) { - case Maxwell::VertexAttribute::Size::Size_16: + case Maxwell::VertexAttribute::Size::Size_R16: return VK_FORMAT_R16_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16: return VK_FORMAT_R16G16_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16: return VK_FORMAT_R16G16B16_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_16_16_16_16: + case Maxwell::VertexAttribute::Size::Size_R16_G16_B16_A16: return VK_FORMAT_R16G16B16A16_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_32: + case Maxwell::VertexAttribute::Size::Size_R32: return VK_FORMAT_R32_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_32_32: + case Maxwell::VertexAttribute::Size::Size_R32_G32: return VK_FORMAT_R32G32_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_32_32_32: + case Maxwell::VertexAttribute::Size::Size_R32_G32_B32: return VK_FORMAT_R32G32B32_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_32_32_32_32: + case Maxwell::VertexAttribute::Size::Size_R32_G32_B32_A32: return VK_FORMAT_R32G32B32A32_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_11_11_10: + case Maxwell::VertexAttribute::Size::Size_B10_G11_R11: return VK_FORMAT_B10G11R11_UFLOAT_PACK32; default: break; @@ -519,29 +542,29 @@ VkFormat VertexFormat(const Device& device, Maxwell::VertexAttribute::Type type, VkCompareOp ComparisonOp(Maxwell::ComparisonOp comparison) { switch (comparison) { - case Maxwell::ComparisonOp::Never: - case Maxwell::ComparisonOp::NeverOld: + case Maxwell::ComparisonOp::Never_D3D: + case Maxwell::ComparisonOp::Never_GL: return VK_COMPARE_OP_NEVER; - case Maxwell::ComparisonOp::Less: - case Maxwell::ComparisonOp::LessOld: + case Maxwell::ComparisonOp::Less_D3D: + case Maxwell::ComparisonOp::Less_GL: return VK_COMPARE_OP_LESS; - case Maxwell::ComparisonOp::Equal: - case Maxwell::ComparisonOp::EqualOld: + case Maxwell::ComparisonOp::Equal_D3D: + case Maxwell::ComparisonOp::Equal_GL: return VK_COMPARE_OP_EQUAL; - case Maxwell::ComparisonOp::LessEqual: - case Maxwell::ComparisonOp::LessEqualOld: + case Maxwell::ComparisonOp::LessEqual_D3D: + case Maxwell::ComparisonOp::LessEqual_GL: return VK_COMPARE_OP_LESS_OR_EQUAL; - case Maxwell::ComparisonOp::Greater: - case Maxwell::ComparisonOp::GreaterOld: + case Maxwell::ComparisonOp::Greater_D3D: + case Maxwell::ComparisonOp::Greater_GL: return VK_COMPARE_OP_GREATER; - case Maxwell::ComparisonOp::NotEqual: - case Maxwell::ComparisonOp::NotEqualOld: + case Maxwell::ComparisonOp::NotEqual_D3D: + case Maxwell::ComparisonOp::NotEqual_GL: return VK_COMPARE_OP_NOT_EQUAL; - case Maxwell::ComparisonOp::GreaterEqual: - case Maxwell::ComparisonOp::GreaterEqualOld: + case Maxwell::ComparisonOp::GreaterEqual_D3D: + case Maxwell::ComparisonOp::GreaterEqual_GL: return VK_COMPARE_OP_GREATER_OR_EQUAL; - case Maxwell::ComparisonOp::Always: - case Maxwell::ComparisonOp::AlwaysOld: + case Maxwell::ComparisonOp::Always_D3D: + case Maxwell::ComparisonOp::Always_GL: return VK_COMPARE_OP_ALWAYS; } UNIMPLEMENTED_MSG("Unimplemented comparison op={}", comparison); @@ -561,31 +584,31 @@ VkIndexType IndexFormat(Maxwell::IndexFormat index_format) { return {}; } -VkStencilOp StencilOp(Maxwell::StencilOp stencil_op) { +VkStencilOp StencilOp(Maxwell::StencilOp::Op stencil_op) { switch (stencil_op) { - case Maxwell::StencilOp::Keep: - case Maxwell::StencilOp::KeepOGL: + case Maxwell::StencilOp::Op::Keep_D3D: + case Maxwell::StencilOp::Op::Keep_GL: return VK_STENCIL_OP_KEEP; - case Maxwell::StencilOp::Zero: - case Maxwell::StencilOp::ZeroOGL: + case Maxwell::StencilOp::Op::Zero_D3D: + case Maxwell::StencilOp::Op::Zero_GL: return VK_STENCIL_OP_ZERO; - case Maxwell::StencilOp::Replace: - case Maxwell::StencilOp::ReplaceOGL: + case Maxwell::StencilOp::Op::Replace_D3D: + case Maxwell::StencilOp::Op::Replace_GL: return VK_STENCIL_OP_REPLACE; - case Maxwell::StencilOp::Incr: - case Maxwell::StencilOp::IncrOGL: + case Maxwell::StencilOp::Op::IncrSaturate_D3D: + case Maxwell::StencilOp::Op::IncrSaturate_GL: return VK_STENCIL_OP_INCREMENT_AND_CLAMP; - case Maxwell::StencilOp::Decr: - case Maxwell::StencilOp::DecrOGL: + case Maxwell::StencilOp::Op::DecrSaturate_D3D: + case Maxwell::StencilOp::Op::DecrSaturate_GL: return VK_STENCIL_OP_DECREMENT_AND_CLAMP; - case Maxwell::StencilOp::Invert: - case Maxwell::StencilOp::InvertOGL: + case Maxwell::StencilOp::Op::Invert_D3D: + case Maxwell::StencilOp::Op::Invert_GL: return VK_STENCIL_OP_INVERT; - case Maxwell::StencilOp::IncrWrap: - case Maxwell::StencilOp::IncrWrapOGL: + case Maxwell::StencilOp::Op::Incr_D3D: + case Maxwell::StencilOp::Op::Incr_GL: return VK_STENCIL_OP_INCREMENT_AND_WRAP; - case Maxwell::StencilOp::DecrWrap: - case Maxwell::StencilOp::DecrWrapOGL: + case Maxwell::StencilOp::Op::Decr_D3D: + case Maxwell::StencilOp::Op::Decr_GL: return VK_STENCIL_OP_DECREMENT_AND_WRAP; } UNIMPLEMENTED_MSG("Unimplemented stencil op={}", stencil_op); @@ -594,20 +617,20 @@ VkStencilOp StencilOp(Maxwell::StencilOp stencil_op) { VkBlendOp BlendEquation(Maxwell::Blend::Equation equation) { switch (equation) { - case Maxwell::Blend::Equation::Add: - case Maxwell::Blend::Equation::AddGL: + case Maxwell::Blend::Equation::Add_D3D: + case Maxwell::Blend::Equation::Add_GL: return VK_BLEND_OP_ADD; - case Maxwell::Blend::Equation::Subtract: - case Maxwell::Blend::Equation::SubtractGL: + case Maxwell::Blend::Equation::Subtract_D3D: + case Maxwell::Blend::Equation::Subtract_GL: return VK_BLEND_OP_SUBTRACT; - case Maxwell::Blend::Equation::ReverseSubtract: - case Maxwell::Blend::Equation::ReverseSubtractGL: + case Maxwell::Blend::Equation::ReverseSubtract_D3D: + case Maxwell::Blend::Equation::ReverseSubtract_GL: return VK_BLEND_OP_REVERSE_SUBTRACT; - case Maxwell::Blend::Equation::Min: - case Maxwell::Blend::Equation::MinGL: + case Maxwell::Blend::Equation::Min_D3D: + case Maxwell::Blend::Equation::Min_GL: return VK_BLEND_OP_MIN; - case Maxwell::Blend::Equation::Max: - case Maxwell::Blend::Equation::MaxGL: + case Maxwell::Blend::Equation::Max_D3D: + case Maxwell::Blend::Equation::Max_GL: return VK_BLEND_OP_MAX; } UNIMPLEMENTED_MSG("Unimplemented blend equation={}", equation); @@ -616,62 +639,62 @@ VkBlendOp BlendEquation(Maxwell::Blend::Equation equation) { VkBlendFactor BlendFactor(Maxwell::Blend::Factor factor) { switch (factor) { - case Maxwell::Blend::Factor::Zero: - case Maxwell::Blend::Factor::ZeroGL: + case Maxwell::Blend::Factor::Zero_D3D: + case Maxwell::Blend::Factor::Zero_GL: return VK_BLEND_FACTOR_ZERO; - case Maxwell::Blend::Factor::One: - case Maxwell::Blend::Factor::OneGL: + case Maxwell::Blend::Factor::One_D3D: + case Maxwell::Blend::Factor::One_GL: return VK_BLEND_FACTOR_ONE; - case Maxwell::Blend::Factor::SourceColor: - case Maxwell::Blend::Factor::SourceColorGL: + case Maxwell::Blend::Factor::SourceColor_D3D: + case Maxwell::Blend::Factor::SourceColor_GL: return VK_BLEND_FACTOR_SRC_COLOR; - case Maxwell::Blend::Factor::OneMinusSourceColor: - case Maxwell::Blend::Factor::OneMinusSourceColorGL: + case Maxwell::Blend::Factor::OneMinusSourceColor_D3D: + case Maxwell::Blend::Factor::OneMinusSourceColor_GL: return VK_BLEND_FACTOR_ONE_MINUS_SRC_COLOR; - case Maxwell::Blend::Factor::SourceAlpha: - case Maxwell::Blend::Factor::SourceAlphaGL: + case Maxwell::Blend::Factor::SourceAlpha_D3D: + case Maxwell::Blend::Factor::SourceAlpha_GL: return VK_BLEND_FACTOR_SRC_ALPHA; - case Maxwell::Blend::Factor::OneMinusSourceAlpha: - case Maxwell::Blend::Factor::OneMinusSourceAlphaGL: + case Maxwell::Blend::Factor::OneMinusSourceAlpha_D3D: + case Maxwell::Blend::Factor::OneMinusSourceAlpha_GL: return VK_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA; - case Maxwell::Blend::Factor::DestAlpha: - case Maxwell::Blend::Factor::DestAlphaGL: + case Maxwell::Blend::Factor::DestAlpha_D3D: + case Maxwell::Blend::Factor::DestAlpha_GL: return VK_BLEND_FACTOR_DST_ALPHA; - case Maxwell::Blend::Factor::OneMinusDestAlpha: - case Maxwell::Blend::Factor::OneMinusDestAlphaGL: + case Maxwell::Blend::Factor::OneMinusDestAlpha_D3D: + case Maxwell::Blend::Factor::OneMinusDestAlpha_GL: return VK_BLEND_FACTOR_ONE_MINUS_DST_ALPHA; - case Maxwell::Blend::Factor::DestColor: - case Maxwell::Blend::Factor::DestColorGL: + case Maxwell::Blend::Factor::DestColor_D3D: + case Maxwell::Blend::Factor::DestColor_GL: return VK_BLEND_FACTOR_DST_COLOR; - case Maxwell::Blend::Factor::OneMinusDestColor: - case Maxwell::Blend::Factor::OneMinusDestColorGL: + case Maxwell::Blend::Factor::OneMinusDestColor_D3D: + case Maxwell::Blend::Factor::OneMinusDestColor_GL: return VK_BLEND_FACTOR_ONE_MINUS_DST_COLOR; - case Maxwell::Blend::Factor::SourceAlphaSaturate: - case Maxwell::Blend::Factor::SourceAlphaSaturateGL: + case Maxwell::Blend::Factor::SourceAlphaSaturate_D3D: + case Maxwell::Blend::Factor::SourceAlphaSaturate_GL: return VK_BLEND_FACTOR_SRC_ALPHA_SATURATE; - case Maxwell::Blend::Factor::Source1Color: - case Maxwell::Blend::Factor::Source1ColorGL: + case Maxwell::Blend::Factor::Source1Color_D3D: + case Maxwell::Blend::Factor::Source1Color_GL: return VK_BLEND_FACTOR_SRC1_COLOR; - case Maxwell::Blend::Factor::OneMinusSource1Color: - case Maxwell::Blend::Factor::OneMinusSource1ColorGL: + case Maxwell::Blend::Factor::OneMinusSource1Color_D3D: + case Maxwell::Blend::Factor::OneMinusSource1Color_GL: return VK_BLEND_FACTOR_ONE_MINUS_SRC1_COLOR; - case Maxwell::Blend::Factor::Source1Alpha: - case Maxwell::Blend::Factor::Source1AlphaGL: + case Maxwell::Blend::Factor::Source1Alpha_D3D: + case Maxwell::Blend::Factor::Source1Alpha_GL: return VK_BLEND_FACTOR_SRC1_ALPHA; - case Maxwell::Blend::Factor::OneMinusSource1Alpha: - case Maxwell::Blend::Factor::OneMinusSource1AlphaGL: + case Maxwell::Blend::Factor::OneMinusSource1Alpha_D3D: + case Maxwell::Blend::Factor::OneMinusSource1Alpha_GL: return VK_BLEND_FACTOR_ONE_MINUS_SRC1_ALPHA; - case Maxwell::Blend::Factor::ConstantColor: - case Maxwell::Blend::Factor::ConstantColorGL: + case Maxwell::Blend::Factor::BlendFactor_D3D: + case Maxwell::Blend::Factor::ConstantColor_GL: return VK_BLEND_FACTOR_CONSTANT_COLOR; - case Maxwell::Blend::Factor::OneMinusConstantColor: - case Maxwell::Blend::Factor::OneMinusConstantColorGL: + case Maxwell::Blend::Factor::OneMinusBlendFactor_D3D: + case Maxwell::Blend::Factor::OneMinusConstantColor_GL: return VK_BLEND_FACTOR_ONE_MINUS_CONSTANT_COLOR; - case Maxwell::Blend::Factor::ConstantAlpha: - case Maxwell::Blend::Factor::ConstantAlphaGL: + case Maxwell::Blend::Factor::BothSourceAlpha_D3D: + case Maxwell::Blend::Factor::ConstantAlpha_GL: return VK_BLEND_FACTOR_CONSTANT_ALPHA; - case Maxwell::Blend::Factor::OneMinusConstantAlpha: - case Maxwell::Blend::Factor::OneMinusConstantAlphaGL: + case Maxwell::Blend::Factor::OneMinusBothSourceAlpha_D3D: + case Maxwell::Blend::Factor::OneMinusConstantAlpha_GL: return VK_BLEND_FACTOR_ONE_MINUS_CONSTANT_ALPHA; } UNIMPLEMENTED_MSG("Unimplemented blend factor={}", factor); diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.h b/src/video_core/renderer_vulkan/maxwell_to_vk.h index 356d46292..6f65502d6 100644 --- a/src/video_core/renderer_vulkan/maxwell_to_vk.h +++ b/src/video_core/renderer_vulkan/maxwell_to_vk.h @@ -55,7 +55,7 @@ VkCompareOp ComparisonOp(Maxwell::ComparisonOp comparison); VkIndexType IndexFormat(Maxwell::IndexFormat index_format); -VkStencilOp StencilOp(Maxwell::StencilOp stencil_op); +VkStencilOp StencilOp(Maxwell::StencilOp::Op stencil_op); VkBlendOp BlendEquation(Maxwell::Blend::Equation equation); diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.cpp b/src/video_core/renderer_vulkan/renderer_vulkan.cpp index 7c78d0299..d8131232a 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.cpp +++ b/src/video_core/renderer_vulkan/renderer_vulkan.cpp @@ -102,13 +102,13 @@ RendererVulkan::RendererVulkan(Core::TelemetrySession& telemetry_session_, debug_callback(Settings::values.renderer_debug ? CreateDebugCallback(instance) : nullptr), surface(CreateSurface(instance, render_window)), device(CreateDevice(instance, dld, *surface)), memory_allocator(device, false), - state_tracker(gpu), scheduler(device, state_tracker), + state_tracker(), scheduler(device, state_tracker), swapchain(*surface, device, scheduler, render_window.GetFramebufferLayout().width, render_window.GetFramebufferLayout().height, false), blit_screen(cpu_memory, render_window, device, memory_allocator, swapchain, scheduler, screen_info), - rasterizer(render_window, gpu, gpu.MemoryManager(), cpu_memory, screen_info, device, - memory_allocator, state_tracker, scheduler) { + rasterizer(render_window, gpu, cpu_memory, screen_info, device, memory_allocator, + state_tracker, scheduler) { Report(); } catch (const vk::Exception& exception) { LOG_ERROR(Render_Vulkan, "Vulkan initialization failed with error: {}", exception.what()); @@ -142,7 +142,7 @@ void RendererVulkan::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) { const auto recreate_swapchain = [&] { if (!has_been_recreated) { has_been_recreated = true; - scheduler.WaitWorker(); + scheduler.Finish(); } const Layout::FramebufferLayout layout = render_window.GetFramebufferLayout(); swapchain.Create(layout.width, layout.height, is_srgb); diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.cpp b/src/video_core/renderer_vulkan/vk_blit_screen.cpp index 444c29f68..89426121f 100644 --- a/src/video_core/renderer_vulkan/vk_blit_screen.cpp +++ b/src/video_core/renderer_vulkan/vk_blit_screen.cpp @@ -145,6 +145,11 @@ VkSemaphore BlitScreen::Draw(const Tegra::FramebufferConfig& framebuffer, // Finish any pending renderpass scheduler.RequestOutsideRenderPassOperationContext(); + if (const auto swapchain_images = swapchain.GetImageCount(); swapchain_images != image_count) { + image_count = swapchain_images; + Recreate(); + } + const std::size_t image_index = swapchain.GetImageIndex(); scheduler.Wait(resource_ticks[image_index]); @@ -448,15 +453,15 @@ vk::Framebuffer BlitScreen::CreateFramebuffer(const VkImageView& image_view, VkE void BlitScreen::CreateStaticResources() { CreateShaders(); + CreateSampler(); +} + +void BlitScreen::CreateDynamicResources() { CreateSemaphores(); CreateDescriptorPool(); CreateDescriptorSetLayout(); CreateDescriptorSets(); CreatePipelineLayout(); - CreateSampler(); -} - -void BlitScreen::CreateDynamicResources() { CreateRenderPass(); CreateFramebuffers(); CreateGraphicsPipeline(); @@ -475,11 +480,15 @@ void BlitScreen::RefreshResources(const Tegra::FramebufferConfig& framebuffer) { fsr.reset(); } - if (framebuffer.width == raw_width && framebuffer.height == raw_height && !raw_images.empty()) { + 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; + ReleaseRawImages(); CreateStagingBuffer(framebuffer); diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.h b/src/video_core/renderer_vulkan/vk_blit_screen.h index b8c67bef0..a2b73ec54 100644 --- a/src/video_core/renderer_vulkan/vk_blit_screen.h +++ b/src/video_core/renderer_vulkan/vk_blit_screen.h @@ -28,6 +28,10 @@ namespace VideoCore { class RasterizerInterface; } +namespace Service::android { +enum class PixelFormat : u32; +} + namespace Vulkan { struct ScreenInfo; @@ -109,7 +113,7 @@ private: MemoryAllocator& memory_allocator; Swapchain& swapchain; Scheduler& scheduler; - const std::size_t image_count; + std::size_t image_count; const ScreenInfo& screen_info; vk::ShaderModule vertex_shader; @@ -156,6 +160,7 @@ private: u32 raw_width = 0; u32 raw_height = 0; + Service::android::PixelFormat pixel_format{}; std::unique_ptr<FSR> fsr; }; diff --git a/src/video_core/renderer_vulkan/vk_compute_pass.cpp b/src/video_core/renderer_vulkan/vk_compute_pass.cpp index f17a5ccd6..241d7573e 100644 --- a/src/video_core/renderer_vulkan/vk_compute_pass.cpp +++ b/src/video_core/renderer_vulkan/vk_compute_pass.cpp @@ -26,8 +26,6 @@ namespace Vulkan { -using Tegra::Texture::SWIZZLE_TABLE; - namespace { constexpr u32 ASTC_BINDING_INPUT_BUFFER = 0; diff --git a/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp b/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp index 6447210e2..7906e11a8 100644 --- a/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp +++ b/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp @@ -126,8 +126,8 @@ void ComputePipeline::Configure(Tegra::Engines::KeplerCompute& kepler_compute, const u32 secondary_offset{desc.secondary_cbuf_offset + index_offset}; const GPUVAddr separate_addr{cbufs[desc.secondary_cbuf_index].Address() + secondary_offset}; - const u32 lhs_raw{gpu_memory.Read<u32>(addr)}; - const u32 rhs_raw{gpu_memory.Read<u32>(separate_addr)}; + const u32 lhs_raw{gpu_memory.Read<u32>(addr) << desc.shift_left}; + const u32 rhs_raw{gpu_memory.Read<u32>(separate_addr) << desc.secondary_shift_left}; return TexturePair(lhs_raw | rhs_raw, via_header_index); } } diff --git a/src/video_core/renderer_vulkan/vk_fence_manager.cpp b/src/video_core/renderer_vulkan/vk_fence_manager.cpp index c249b34d4..0214b103a 100644 --- a/src/video_core/renderer_vulkan/vk_fence_manager.cpp +++ b/src/video_core/renderer_vulkan/vk_fence_manager.cpp @@ -11,11 +11,8 @@ namespace Vulkan { -InnerFence::InnerFence(Scheduler& scheduler_, u32 payload_, bool is_stubbed_) - : FenceBase{payload_, is_stubbed_}, scheduler{scheduler_} {} - -InnerFence::InnerFence(Scheduler& scheduler_, GPUVAddr address_, u32 payload_, bool is_stubbed_) - : FenceBase{address_, payload_, is_stubbed_}, scheduler{scheduler_} {} +InnerFence::InnerFence(Scheduler& scheduler_, bool is_stubbed_) + : FenceBase{is_stubbed_}, scheduler{scheduler_} {} InnerFence::~InnerFence() = default; @@ -48,12 +45,8 @@ FenceManager::FenceManager(VideoCore::RasterizerInterface& rasterizer_, Tegra::G : GenericFenceManager{rasterizer_, gpu_, texture_cache_, buffer_cache_, query_cache_}, scheduler{scheduler_} {} -Fence FenceManager::CreateFence(u32 value, bool is_stubbed) { - return std::make_shared<InnerFence>(scheduler, value, is_stubbed); -} - -Fence FenceManager::CreateFence(GPUVAddr addr, u32 value, bool is_stubbed) { - return std::make_shared<InnerFence>(scheduler, addr, value, is_stubbed); +Fence FenceManager::CreateFence(bool is_stubbed) { + return std::make_shared<InnerFence>(scheduler, is_stubbed); } void FenceManager::QueueFence(Fence& fence) { diff --git a/src/video_core/renderer_vulkan/vk_fence_manager.h b/src/video_core/renderer_vulkan/vk_fence_manager.h index 7c0bbd80a..7fe2afcd9 100644 --- a/src/video_core/renderer_vulkan/vk_fence_manager.h +++ b/src/video_core/renderer_vulkan/vk_fence_manager.h @@ -25,8 +25,7 @@ class Scheduler; class InnerFence : public VideoCommon::FenceBase { public: - explicit InnerFence(Scheduler& scheduler_, u32 payload_, bool is_stubbed_); - explicit InnerFence(Scheduler& scheduler_, GPUVAddr address_, u32 payload_, bool is_stubbed_); + explicit InnerFence(Scheduler& scheduler_, bool is_stubbed_); ~InnerFence(); void Queue(); @@ -50,8 +49,7 @@ public: QueryCache& query_cache, const Device& device, Scheduler& scheduler); protected: - Fence CreateFence(u32 value, bool is_stubbed) override; - Fence CreateFence(GPUVAddr addr, u32 value, bool is_stubbed) override; + Fence CreateFence(bool is_stubbed) override; void QueueFence(Fence& fence) override; bool IsFenceSignaled(Fence& fence) const override; void WaitFence(Fence& fence) override; diff --git a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp index 7eb623a7c..b4372a839 100644 --- a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp +++ b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp @@ -216,15 +216,14 @@ ConfigureFuncPtr ConfigureFunc(const std::array<vk::ShaderModule, NUM_STAGES>& m } // Anonymous namespace GraphicsPipeline::GraphicsPipeline( - Tegra::Engines::Maxwell3D& maxwell3d_, Tegra::MemoryManager& gpu_memory_, Scheduler& scheduler_, - BufferCache& buffer_cache_, TextureCache& texture_cache_, + Scheduler& scheduler_, BufferCache& buffer_cache_, TextureCache& texture_cache_, VideoCore::ShaderNotify* shader_notify, const Device& device_, DescriptorPool& descriptor_pool, UpdateDescriptorQueue& update_descriptor_queue_, Common::ThreadWorker* worker_thread, PipelineStatistics* pipeline_statistics, RenderPassCache& render_pass_cache, const GraphicsPipelineCacheKey& key_, std::array<vk::ShaderModule, NUM_STAGES> stages, const std::array<const Shader::Info*, NUM_STAGES>& infos) - : key{key_}, maxwell3d{maxwell3d_}, gpu_memory{gpu_memory_}, device{device_}, - texture_cache{texture_cache_}, buffer_cache{buffer_cache_}, scheduler{scheduler_}, + : key{key_}, device{device_}, texture_cache{texture_cache_}, + buffer_cache{buffer_cache_}, scheduler{scheduler_}, update_descriptor_queue{update_descriptor_queue_}, spv_modules{std::move(stages)} { if (shader_notify) { shader_notify->MarkShaderBuilding(); @@ -289,8 +288,8 @@ void GraphicsPipeline::ConfigureImpl(bool is_indexed) { buffer_cache.SetUniformBuffersState(enabled_uniform_buffer_masks, &uniform_buffer_sizes); - const auto& regs{maxwell3d.regs}; - const bool via_header_index{regs.sampler_index == Maxwell::SamplerIndex::ViaHeaderIndex}; + const auto& regs{maxwell3d->regs}; + const bool via_header_index{regs.sampler_binding == Maxwell::SamplerBinding::ViaHeaderBinding}; const auto config_stage{[&](size_t stage) LAMBDA_FORCEINLINE { const Shader::Info& info{stage_infos[stage]}; buffer_cache.UnbindGraphicsStorageBuffers(stage); @@ -303,7 +302,7 @@ void GraphicsPipeline::ConfigureImpl(bool is_indexed) { ++ssbo_index; } } - const auto& cbufs{maxwell3d.state.shader_stages[stage].const_buffers}; + const auto& cbufs{maxwell3d->state.shader_stages[stage].const_buffers}; const auto read_handle{[&](const auto& desc, u32 index) { ASSERT(cbufs[desc.cbuf_index].enabled); const u32 index_offset{index << desc.size_shift}; @@ -316,13 +315,14 @@ void GraphicsPipeline::ConfigureImpl(bool is_indexed) { const u32 second_offset{desc.secondary_cbuf_offset + index_offset}; const GPUVAddr separate_addr{cbufs[desc.secondary_cbuf_index].address + second_offset}; - const u32 lhs_raw{gpu_memory.Read<u32>(addr)}; - const u32 rhs_raw{gpu_memory.Read<u32>(separate_addr)}; + const u32 lhs_raw{gpu_memory->Read<u32>(addr) << desc.shift_left}; + const u32 rhs_raw{gpu_memory->Read<u32>(separate_addr) + << desc.secondary_shift_left}; const u32 raw{lhs_raw | rhs_raw}; return TexturePair(raw, via_header_index); } } - return TexturePair(gpu_memory.Read<u32>(addr), via_header_index); + return TexturePair(gpu_memory->Read<u32>(addr), via_header_index); }}; const auto add_image{[&](const auto& desc, bool blacklist) LAMBDA_FORCEINLINE { for (u32 index = 0; index < desc.count; ++index) { @@ -680,15 +680,6 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) { .lineStippleFactor = 0, .lineStipplePattern = 0, }; - VkPipelineRasterizationConservativeStateCreateInfoEXT conservative_raster{ - .sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_CONSERVATIVE_STATE_CREATE_INFO_EXT, - .pNext = nullptr, - .flags = 0, - .conservativeRasterizationMode = key.state.conservative_raster_enable != 0 - ? VK_CONSERVATIVE_RASTERIZATION_MODE_OVERESTIMATE_EXT - : VK_CONSERVATIVE_RASTERIZATION_MODE_DISABLED_EXT, - .extraPrimitiveOverestimationSize = 0.0f, - }; VkPipelineRasterizationProvokingVertexStateCreateInfoEXT provoking_vertex{ .sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_PROVOKING_VERTEX_STATE_CREATE_INFO_EXT, .pNext = nullptr, @@ -699,9 +690,6 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) { if (IsLine(input_assembly_topology) && device.IsExtLineRasterizationSupported()) { line_state.pNext = std::exchange(rasterization_ci.pNext, &line_state); } - if (device.IsExtConservativeRasterizationSupported()) { - conservative_raster.pNext = std::exchange(rasterization_ci.pNext, &conservative_raster); - } if (device.IsExtProvokingVertexSupported()) { provoking_vertex.pNext = std::exchange(rasterization_ci.pNext, &provoking_vertex); } diff --git a/src/video_core/renderer_vulkan/vk_graphics_pipeline.h b/src/video_core/renderer_vulkan/vk_graphics_pipeline.h index 8842f62d7..6bf577d25 100644 --- a/src/video_core/renderer_vulkan/vk_graphics_pipeline.h +++ b/src/video_core/renderer_vulkan/vk_graphics_pipeline.h @@ -70,15 +70,16 @@ class GraphicsPipeline { static constexpr size_t NUM_STAGES = Tegra::Engines::Maxwell3D::Regs::MaxShaderStage; public: - explicit GraphicsPipeline( - Tegra::Engines::Maxwell3D& maxwell3d, Tegra::MemoryManager& gpu_memory, - Scheduler& scheduler, BufferCache& buffer_cache, TextureCache& texture_cache, - VideoCore::ShaderNotify* shader_notify, const Device& device, - DescriptorPool& descriptor_pool, UpdateDescriptorQueue& update_descriptor_queue, - Common::ThreadWorker* worker_thread, PipelineStatistics* pipeline_statistics, - RenderPassCache& render_pass_cache, const GraphicsPipelineCacheKey& key, - std::array<vk::ShaderModule, NUM_STAGES> stages, - const std::array<const Shader::Info*, NUM_STAGES>& infos); + explicit GraphicsPipeline(Scheduler& scheduler, BufferCache& buffer_cache, + TextureCache& texture_cache, VideoCore::ShaderNotify* shader_notify, + const Device& device, DescriptorPool& descriptor_pool, + UpdateDescriptorQueue& update_descriptor_queue, + Common::ThreadWorker* worker_thread, + PipelineStatistics* pipeline_statistics, + RenderPassCache& render_pass_cache, + const GraphicsPipelineCacheKey& key, + std::array<vk::ShaderModule, NUM_STAGES> stages, + const std::array<const Shader::Info*, NUM_STAGES>& infos); GraphicsPipeline& operator=(GraphicsPipeline&&) noexcept = delete; GraphicsPipeline(GraphicsPipeline&&) noexcept = delete; @@ -110,6 +111,11 @@ public: return [](GraphicsPipeline* pl, bool is_indexed) { pl->ConfigureImpl<Spec>(is_indexed); }; } + void SetEngine(Tegra::Engines::Maxwell3D* maxwell3d_, Tegra::MemoryManager* gpu_memory_) { + maxwell3d = maxwell3d_; + gpu_memory = gpu_memory_; + } + private: template <typename Spec> void ConfigureImpl(bool is_indexed); @@ -122,8 +128,8 @@ private: void Validate(); const GraphicsPipelineCacheKey key; - Tegra::Engines::Maxwell3D& maxwell3d; - Tegra::MemoryManager& gpu_memory; + Tegra::Engines::Maxwell3D* maxwell3d; + Tegra::MemoryManager* gpu_memory; const Device& device; TextureCache& texture_cache; BufferCache& buffer_cache; diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp index 9708dc45e..13d5a1f67 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp @@ -62,29 +62,29 @@ auto MakeSpan(Container& container) { Shader::CompareFunction MaxwellToCompareFunction(Maxwell::ComparisonOp comparison) { switch (comparison) { - case Maxwell::ComparisonOp::Never: - case Maxwell::ComparisonOp::NeverOld: + case Maxwell::ComparisonOp::Never_D3D: + case Maxwell::ComparisonOp::Never_GL: return Shader::CompareFunction::Never; - case Maxwell::ComparisonOp::Less: - case Maxwell::ComparisonOp::LessOld: + case Maxwell::ComparisonOp::Less_D3D: + case Maxwell::ComparisonOp::Less_GL: return Shader::CompareFunction::Less; - case Maxwell::ComparisonOp::Equal: - case Maxwell::ComparisonOp::EqualOld: + case Maxwell::ComparisonOp::Equal_D3D: + case Maxwell::ComparisonOp::Equal_GL: return Shader::CompareFunction::Equal; - case Maxwell::ComparisonOp::LessEqual: - case Maxwell::ComparisonOp::LessEqualOld: + case Maxwell::ComparisonOp::LessEqual_D3D: + case Maxwell::ComparisonOp::LessEqual_GL: return Shader::CompareFunction::LessThanEqual; - case Maxwell::ComparisonOp::Greater: - case Maxwell::ComparisonOp::GreaterOld: + case Maxwell::ComparisonOp::Greater_D3D: + case Maxwell::ComparisonOp::Greater_GL: return Shader::CompareFunction::Greater; - case Maxwell::ComparisonOp::NotEqual: - case Maxwell::ComparisonOp::NotEqualOld: + case Maxwell::ComparisonOp::NotEqual_D3D: + case Maxwell::ComparisonOp::NotEqual_GL: return Shader::CompareFunction::NotEqual; - case Maxwell::ComparisonOp::GreaterEqual: - case Maxwell::ComparisonOp::GreaterEqualOld: + case Maxwell::ComparisonOp::GreaterEqual_D3D: + case Maxwell::ComparisonOp::GreaterEqual_GL: return Shader::CompareFunction::GreaterThanEqual; - case Maxwell::ComparisonOp::Always: - case Maxwell::ComparisonOp::AlwaysOld: + case Maxwell::ComparisonOp::Always_D3D: + case Maxwell::ComparisonOp::Always_GL: return Shader::CompareFunction::Always; } UNIMPLEMENTED_MSG("Unimplemented comparison op={}", comparison); @@ -96,15 +96,18 @@ Shader::AttributeType CastAttributeType(const FixedPipelineState::VertexAttribut return Shader::AttributeType::Disabled; } switch (attr.Type()) { - case Maxwell::VertexAttribute::Type::SignedNorm: - case Maxwell::VertexAttribute::Type::UnsignedNorm: - case Maxwell::VertexAttribute::Type::UnsignedScaled: - case Maxwell::VertexAttribute::Type::SignedScaled: + case Maxwell::VertexAttribute::Type::UnusedEnumDoNotUseBecauseItWillGoAway: + ASSERT_MSG(false, "Invalid vertex attribute type!"); + return Shader::AttributeType::Disabled; + case Maxwell::VertexAttribute::Type::SNorm: + case Maxwell::VertexAttribute::Type::UNorm: + case Maxwell::VertexAttribute::Type::UScaled: + case Maxwell::VertexAttribute::Type::SScaled: case Maxwell::VertexAttribute::Type::Float: return Shader::AttributeType::Float; - case Maxwell::VertexAttribute::Type::SignedInt: + case Maxwell::VertexAttribute::Type::SInt: return Shader::AttributeType::SignedInt; - case Maxwell::VertexAttribute::Type::UnsignedInt: + case Maxwell::VertexAttribute::Type::UInt: return Shader::AttributeType::UnsignedInt; } return Shader::AttributeType::Float; @@ -131,6 +134,7 @@ Shader::RuntimeInfo MakeRuntimeInfo(std::span<const Shader::IR::Program> program Shader::RuntimeInfo info; if (previous_program) { info.previous_stage_stores = previous_program->info.stores; + info.previous_stage_legacy_stores_mapping = previous_program->info.legacy_stores_mapping; if (previous_program->is_geometry_passthrough) { info.previous_stage_stores.mask |= previous_program->info.passthrough.mask; } @@ -162,16 +166,14 @@ Shader::RuntimeInfo MakeRuntimeInfo(std::span<const Shader::IR::Program> program } break; case Shader::Stage::TessellationEval: - // We have to flip tessellation clockwise for some reason... - info.tess_clockwise = key.state.tessellation_clockwise == 0; info.tess_primitive = [&key] { const u32 raw{key.state.tessellation_primitive.Value()}; - switch (static_cast<Maxwell::TessellationPrimitive>(raw)) { - case Maxwell::TessellationPrimitive::Isolines: + switch (static_cast<Maxwell::Tessellation::DomainType>(raw)) { + case Maxwell::Tessellation::DomainType::Isolines: return Shader::TessPrimitive::Isolines; - case Maxwell::TessellationPrimitive::Triangles: + case Maxwell::Tessellation::DomainType::Triangles: return Shader::TessPrimitive::Triangles; - case Maxwell::TessellationPrimitive::Quads: + case Maxwell::Tessellation::DomainType::Quads: return Shader::TessPrimitive::Quads; } ASSERT(false); @@ -179,12 +181,12 @@ Shader::RuntimeInfo MakeRuntimeInfo(std::span<const Shader::IR::Program> program }(); info.tess_spacing = [&] { const u32 raw{key.state.tessellation_spacing}; - switch (static_cast<Maxwell::TessellationSpacing>(raw)) { - case Maxwell::TessellationSpacing::Equal: + switch (static_cast<Maxwell::Tessellation::Spacing>(raw)) { + case Maxwell::Tessellation::Spacing::Integer: return Shader::TessSpacing::Equal; - case Maxwell::TessellationSpacing::FractionalOdd: + case Maxwell::Tessellation::Spacing::FractionalOdd: return Shader::TessSpacing::FractionalOdd; - case Maxwell::TessellationSpacing::FractionalEven: + case Maxwell::Tessellation::Spacing::FractionalEven: return Shader::TessSpacing::FractionalEven; } ASSERT(false); @@ -259,20 +261,18 @@ bool GraphicsPipelineCacheKey::operator==(const GraphicsPipelineCacheKey& rhs) c return std::memcmp(&rhs, this, Size()) == 0; } -PipelineCache::PipelineCache(RasterizerVulkan& rasterizer_, Tegra::Engines::Maxwell3D& maxwell3d_, - Tegra::Engines::KeplerCompute& kepler_compute_, - Tegra::MemoryManager& gpu_memory_, const Device& device_, +PipelineCache::PipelineCache(RasterizerVulkan& rasterizer_, const Device& device_, Scheduler& scheduler_, DescriptorPool& descriptor_pool_, UpdateDescriptorQueue& update_descriptor_queue_, RenderPassCache& render_pass_cache_, BufferCache& buffer_cache_, TextureCache& texture_cache_, VideoCore::ShaderNotify& shader_notify_) - : VideoCommon::ShaderCache{rasterizer_, gpu_memory_, maxwell3d_, kepler_compute_}, - device{device_}, scheduler{scheduler_}, descriptor_pool{descriptor_pool_}, - update_descriptor_queue{update_descriptor_queue_}, render_pass_cache{render_pass_cache_}, - buffer_cache{buffer_cache_}, texture_cache{texture_cache_}, shader_notify{shader_notify_}, + : VideoCommon::ShaderCache{rasterizer_}, device{device_}, scheduler{scheduler_}, + descriptor_pool{descriptor_pool_}, update_descriptor_queue{update_descriptor_queue_}, + render_pass_cache{render_pass_cache_}, buffer_cache{buffer_cache_}, + texture_cache{texture_cache_}, shader_notify{shader_notify_}, use_asynchronous_shaders{Settings::values.use_asynchronous_shaders.GetValue()}, - workers(std::max(std::thread::hardware_concurrency(), 2U) - 1, "yuzu:PipelineBuilder"), - serialization_thread(1, "yuzu:PipelineSerialization") { + workers(std::max(std::thread::hardware_concurrency(), 2U) - 1, "VkPipelineBuilder"), + serialization_thread(1, "VkPipelineSerialization") { const auto& float_control{device.FloatControlProperties()}; const VkDriverIdKHR driver_id{device.GetDriverID()}; profile = Shader::Profile{ @@ -337,7 +337,7 @@ GraphicsPipeline* PipelineCache::CurrentGraphicsPipeline() { current_pipeline = nullptr; return nullptr; } - graphics_key.state.Refresh(maxwell3d, device.IsExtExtendedDynamicStateSupported(), + graphics_key.state.Refresh(*maxwell3d, device.IsExtExtendedDynamicStateSupported(), device.IsExtVertexInputDynamicStateSupported()); if (current_pipeline) { @@ -357,7 +357,7 @@ ComputePipeline* PipelineCache::CurrentComputePipeline() { if (!shader) { return nullptr; } - const auto& qmd{kepler_compute.launch_description}; + const auto& qmd{kepler_compute->launch_description}; const ComputePipelineCacheKey key{ .unique_hash = shader->unique_hash, .shared_memory_size = qmd.shared_alloc, @@ -486,13 +486,13 @@ GraphicsPipeline* PipelineCache::BuiltPipeline(GraphicsPipeline* pipeline) const } // If something is using depth, we can assume that games are not rendering anything which // will be used one time. - if (maxwell3d.regs.zeta_enable) { + if (maxwell3d->regs.zeta_enable) { return nullptr; } // If games are using a small index count, we can assume these are full screen quads. // Usually these shaders are only used once for building textures so we can assume they // can't be built async - if (maxwell3d.regs.index_array.count <= 6 || maxwell3d.regs.vertex_buffer.count <= 6) { + if (maxwell3d->regs.index_buffer.count <= 6 || maxwell3d->regs.vertex_buffer.count <= 6) { return pipeline; } return nullptr; @@ -557,10 +557,10 @@ std::unique_ptr<GraphicsPipeline> PipelineCache::CreateGraphicsPipeline( previous_stage = &program; } Common::ThreadWorker* const thread_worker{build_in_parallel ? &workers : nullptr}; - return std::make_unique<GraphicsPipeline>( - maxwell3d, gpu_memory, scheduler, buffer_cache, texture_cache, &shader_notify, device, - descriptor_pool, update_descriptor_queue, thread_worker, statistics, render_pass_cache, key, - std::move(modules), infos); + return std::make_unique<GraphicsPipeline>(scheduler, buffer_cache, texture_cache, + &shader_notify, device, descriptor_pool, + update_descriptor_queue, thread_worker, statistics, + render_pass_cache, key, std::move(modules), infos); } catch (const Shader::Exception& exception) { LOG_ERROR(Render_Vulkan, "{}", exception.what()); @@ -592,9 +592,9 @@ std::unique_ptr<GraphicsPipeline> PipelineCache::CreateGraphicsPipeline() { std::unique_ptr<ComputePipeline> PipelineCache::CreateComputePipeline( const ComputePipelineCacheKey& key, const ShaderInfo* shader) { - const GPUVAddr program_base{kepler_compute.regs.code_loc.Address()}; - const auto& qmd{kepler_compute.launch_description}; - ComputeEnvironment env{kepler_compute, gpu_memory, program_base, qmd.program_start}; + const GPUVAddr program_base{kepler_compute->regs.code_loc.Address()}; + const auto& qmd{kepler_compute->launch_description}; + ComputeEnvironment env{*kepler_compute, *gpu_memory, program_base, qmd.program_start}; env.SetCachedSize(shader->size_bytes); main_pools.ReleaseContents(); diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.h b/src/video_core/renderer_vulkan/vk_pipeline_cache.h index 127957dbf..61f9e9366 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.h +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.h @@ -100,10 +100,8 @@ struct ShaderPools { class PipelineCache : public VideoCommon::ShaderCache { public: - explicit PipelineCache(RasterizerVulkan& rasterizer, Tegra::Engines::Maxwell3D& maxwell3d, - Tegra::Engines::KeplerCompute& kepler_compute, - Tegra::MemoryManager& gpu_memory, const Device& device, - Scheduler& scheduler, DescriptorPool& descriptor_pool, + explicit PipelineCache(RasterizerVulkan& rasterizer, const Device& device, Scheduler& scheduler, + DescriptorPool& descriptor_pool, UpdateDescriptorQueue& update_descriptor_queue, RenderPassCache& render_pass_cache, BufferCache& buffer_cache, TextureCache& texture_cache, VideoCore::ShaderNotify& shader_notify_); diff --git a/src/video_core/renderer_vulkan/vk_query_cache.cpp b/src/video_core/renderer_vulkan/vk_query_cache.cpp index 2b859c6b8..4b15c0f85 100644 --- a/src/video_core/renderer_vulkan/vk_query_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_query_cache.cpp @@ -59,16 +59,16 @@ void QueryPool::Reserve(std::pair<VkQueryPool, u32> query) { std::find_if(pools.begin(), pools.end(), [query_pool = query.first](vk::QueryPool& pool) { return query_pool == *pool; }); - ASSERT(it != std::end(pools)); - const std::ptrdiff_t pool_index = std::distance(std::begin(pools), it); - usage[pool_index * GROW_STEP + static_cast<std::ptrdiff_t>(query.second)] = false; + if (it != std::end(pools)) { + const std::ptrdiff_t pool_index = std::distance(std::begin(pools), it); + usage[pool_index * GROW_STEP + static_cast<std::ptrdiff_t>(query.second)] = false; + } } -QueryCache::QueryCache(VideoCore::RasterizerInterface& rasterizer_, - Tegra::Engines::Maxwell3D& maxwell3d_, Tegra::MemoryManager& gpu_memory_, - const Device& device_, Scheduler& scheduler_) - : QueryCacheBase{rasterizer_, maxwell3d_, gpu_memory_}, device{device_}, scheduler{scheduler_}, +QueryCache::QueryCache(VideoCore::RasterizerInterface& rasterizer_, const Device& device_, + Scheduler& scheduler_) + : QueryCacheBase{rasterizer_}, device{device_}, scheduler{scheduler_}, query_pools{ QueryPool{device_, scheduler_, QueryType::SamplesPassed}, } {} diff --git a/src/video_core/renderer_vulkan/vk_query_cache.h b/src/video_core/renderer_vulkan/vk_query_cache.h index b0d86c4f8..26762ee09 100644 --- a/src/video_core/renderer_vulkan/vk_query_cache.h +++ b/src/video_core/renderer_vulkan/vk_query_cache.h @@ -52,9 +52,8 @@ private: class QueryCache final : public VideoCommon::QueryCacheBase<QueryCache, CachedQuery, CounterStream, HostCounter> { public: - explicit QueryCache(VideoCore::RasterizerInterface& rasterizer_, - Tegra::Engines::Maxwell3D& maxwell3d_, Tegra::MemoryManager& gpu_memory_, - const Device& device_, Scheduler& scheduler_); + explicit QueryCache(VideoCore::RasterizerInterface& rasterizer_, const Device& device_, + Scheduler& scheduler_); ~QueryCache(); std::pair<VkQueryPool, u32> AllocateQuery(VideoCore::QueryType type); diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index bf9f825f4..5af3c930b 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -11,6 +11,7 @@ #include "common/microprofile.h" #include "common/scope_exit.h" #include "common/settings.h" +#include "video_core/control/channel_state.h" #include "video_core/engines/kepler_compute.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/renderer_vulkan/blit_image.h" @@ -69,7 +70,7 @@ VkViewport GetViewportState(const Device& device, const Maxwell& regs, size_t in const float width = conv(src.scale_x * 2.0f); float y = conv(src.translate_y - src.scale_y); float height = conv(src.scale_y * 2.0f); - bool y_negate = regs.screen_y_control.y_negate; + bool y_negate = regs.window_origin.mode != Maxwell::WindowOrigin::Mode::UpperLeft; if (!device.IsNvViewportSwizzleSupported()) { y_negate = y_negate != (src.swizzle.y == Maxwell::ViewportSwizzle::NegativeY); @@ -126,14 +127,13 @@ VkRect2D GetScissorState(const Maxwell& regs, size_t index, u32 up_scale = 1, u3 return scissor; } -DrawParams MakeDrawParams(const Maxwell& regs, u32 num_instances, bool is_instanced, - bool is_indexed) { +DrawParams MakeDrawParams(const Maxwell& regs, u32 num_instances, bool is_indexed) { DrawParams params{ - .base_instance = regs.vb_base_instance, - .num_instances = is_instanced ? num_instances : 1, - .base_vertex = is_indexed ? regs.vb_element_base : regs.vertex_buffer.first, - .num_vertices = is_indexed ? regs.index_array.count : regs.vertex_buffer.count, - .first_index = is_indexed ? regs.index_array.first : 0, + .base_instance = regs.global_base_instance_index, + .num_instances = num_instances, + .base_vertex = is_indexed ? regs.global_base_vertex_index : regs.vertex_buffer.first, + .num_vertices = is_indexed ? regs.index_buffer.count : regs.vertex_buffer.count, + .first_index = is_indexed ? regs.index_buffer.first : 0, .is_indexed = is_indexed, }; if (regs.draw.topology == Maxwell::PrimitiveTopology::Quads) { @@ -148,31 +148,25 @@ DrawParams MakeDrawParams(const Maxwell& regs, u32 num_instances, bool is_instan } // Anonymous namespace RasterizerVulkan::RasterizerVulkan(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_, - Tegra::MemoryManager& gpu_memory_, Core::Memory::Memory& cpu_memory_, ScreenInfo& screen_info_, const Device& device_, MemoryAllocator& memory_allocator_, StateTracker& state_tracker_, Scheduler& scheduler_) - : RasterizerAccelerated{cpu_memory_}, gpu{gpu_}, - gpu_memory{gpu_memory_}, maxwell3d{gpu.Maxwell3D()}, kepler_compute{gpu.KeplerCompute()}, - screen_info{screen_info_}, device{device_}, memory_allocator{memory_allocator_}, - state_tracker{state_tracker_}, scheduler{scheduler_}, + : RasterizerAccelerated{cpu_memory_}, gpu{gpu_}, screen_info{screen_info_}, device{device_}, + memory_allocator{memory_allocator_}, state_tracker{state_tracker_}, scheduler{scheduler_}, staging_pool(device, memory_allocator, scheduler), descriptor_pool(device, scheduler), update_descriptor_queue(device, scheduler), blit_image(device, scheduler, state_tracker, descriptor_pool), - astc_decoder_pass(device, scheduler, descriptor_pool, staging_pool, update_descriptor_queue, - memory_allocator), render_pass_cache(device), texture_cache_runtime{device, scheduler, memory_allocator, staging_pool, - blit_image, astc_decoder_pass, - render_pass_cache}, - texture_cache(texture_cache_runtime, *this, maxwell3d, kepler_compute, gpu_memory), + blit_image, render_pass_cache, + descriptor_pool, update_descriptor_queue}, + texture_cache(texture_cache_runtime, *this), buffer_cache_runtime(device, memory_allocator, scheduler, staging_pool, update_descriptor_queue, descriptor_pool), - buffer_cache(*this, maxwell3d, kepler_compute, gpu_memory, cpu_memory_, buffer_cache_runtime), - pipeline_cache(*this, maxwell3d, kepler_compute, gpu_memory, device, scheduler, - descriptor_pool, update_descriptor_queue, render_pass_cache, buffer_cache, - texture_cache, gpu.ShaderNotify()), - query_cache{*this, maxwell3d, gpu_memory, device, scheduler}, accelerate_dma{buffer_cache}, + buffer_cache(*this, cpu_memory_, buffer_cache_runtime), + pipeline_cache(*this, device, scheduler, descriptor_pool, update_descriptor_queue, + render_pass_cache, buffer_cache, texture_cache, gpu.ShaderNotify()), + query_cache{*this, device, scheduler}, accelerate_dma{buffer_cache}, fence_manager(*this, gpu, texture_cache, buffer_cache, query_cache, device, scheduler), wfi_event(device.GetLogical().CreateEvent()) { scheduler.SetQueryCache(query_cache); @@ -180,7 +174,7 @@ RasterizerVulkan::RasterizerVulkan(Core::Frontend::EmuWindow& emu_window_, Tegra RasterizerVulkan::~RasterizerVulkan() = default; -void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) { +void RasterizerVulkan::Draw(bool is_indexed, u32 instance_count) { MICROPROFILE_SCOPE(Vulkan_Drawing); SCOPE_EXIT({ gpu.TickWork(); }); @@ -193,15 +187,19 @@ void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) { return; } std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex}; + // update engine as channel may be different. + pipeline->SetEngine(maxwell3d, gpu_memory); pipeline->Configure(is_indexed); + BindInlineIndexBuffer(); + BeginTransformFeedback(); UpdateDynamicStates(); - const auto& regs{maxwell3d.regs}; - const u32 num_instances{maxwell3d.mme_draw.instance_count}; - const DrawParams draw_params{MakeDrawParams(regs, num_instances, is_instanced, is_indexed)}; + const auto& regs{maxwell3d->regs}; + const u32 num_instances{instance_count}; + const DrawParams draw_params{MakeDrawParams(regs, num_instances, is_indexed)}; scheduler.Record([draw_params](vk::CommandBuffer cmdbuf) { if (draw_params.is_indexed) { cmdbuf.DrawIndexed(draw_params.num_vertices, draw_params.num_instances, @@ -218,18 +216,18 @@ void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) { void RasterizerVulkan::Clear() { MICROPROFILE_SCOPE(Vulkan_Clearing); - if (!maxwell3d.ShouldExecute()) { + if (!maxwell3d->ShouldExecute()) { return; } FlushWork(); query_cache.UpdateCounters(); - auto& regs = maxwell3d.regs; - const bool use_color = regs.clear_buffers.R || regs.clear_buffers.G || regs.clear_buffers.B || - regs.clear_buffers.A; - const bool use_depth = regs.clear_buffers.Z; - const bool use_stencil = regs.clear_buffers.S; + auto& regs = maxwell3d->regs; + const bool use_color = regs.clear_surface.R || regs.clear_surface.G || regs.clear_surface.B || + regs.clear_surface.A; + const bool use_depth = regs.clear_surface.Z; + const bool use_stencil = regs.clear_surface.S; if (!use_color && !use_depth && !use_stencil) { return; } @@ -248,9 +246,16 @@ void RasterizerVulkan::Clear() { } UpdateViewportsState(regs); + VkRect2D default_scissor; + default_scissor.offset.x = 0; + default_scissor.offset.y = 0; + default_scissor.extent.width = std::numeric_limits<s32>::max(); + default_scissor.extent.height = std::numeric_limits<s32>::max(); + VkClearRect clear_rect{ - .rect = GetScissorState(regs, 0, up_scale, down_shift), - .baseArrayLayer = regs.clear_buffers.layer, + .rect = regs.clear_control.use_scissor ? GetScissorState(regs, 0, up_scale, down_shift) + : default_scissor, + .baseArrayLayer = regs.clear_surface.layer, .layerCount = 1, }; if (clear_rect.rect.extent.width == 0 || clear_rect.rect.extent.height == 0) { @@ -261,7 +266,7 @@ void RasterizerVulkan::Clear() { .height = std::min(clear_rect.rect.extent.height, render_area.height), }; - const u32 color_attachment = regs.clear_buffers.RT; + const u32 color_attachment = regs.clear_surface.RT; if (use_color && framebuffer->HasAspectColorBit(color_attachment)) { VkClearValue clear_value; bool is_integer = false; @@ -283,7 +288,8 @@ void RasterizerVulkan::Clear() { break; } if (!is_integer) { - std::memcpy(clear_value.color.float32, regs.clear_color, sizeof(regs.clear_color)); + std::memcpy(clear_value.color.float32, regs.clear_color.data(), + regs.clear_color.size() * sizeof(f32)); } else if (!is_signed) { for (size_t i = 0; i < 4; i++) { clear_value.color.uint32[i] = static_cast<u32>( @@ -297,14 +303,19 @@ void RasterizerVulkan::Clear() { } } - scheduler.Record([color_attachment, clear_value, clear_rect](vk::CommandBuffer cmdbuf) { - const VkClearAttachment attachment{ - .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, - .colorAttachment = color_attachment, - .clearValue = clear_value, - }; - cmdbuf.ClearAttachments(attachment, clear_rect); - }); + if (regs.clear_surface.R && regs.clear_surface.G && regs.clear_surface.B && + regs.clear_surface.A) { + scheduler.Record([color_attachment, clear_value, clear_rect](vk::CommandBuffer cmdbuf) { + const VkClearAttachment attachment{ + .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT, + .colorAttachment = color_attachment, + .clearValue = clear_value, + }; + cmdbuf.ClearAttachments(attachment, clear_rect); + }); + } else { + UNIMPLEMENTED_MSG("Unimplemented Clear only the specified channel"); + } } if (!use_depth && !use_stencil) { @@ -339,9 +350,9 @@ void RasterizerVulkan::DispatchCompute() { return; } std::scoped_lock lock{texture_cache.mutex, buffer_cache.mutex}; - pipeline->Configure(kepler_compute, gpu_memory, scheduler, buffer_cache, texture_cache); + pipeline->Configure(*kepler_compute, *gpu_memory, scheduler, buffer_cache, texture_cache); - const auto& qmd{kepler_compute.launch_description}; + const auto& qmd{kepler_compute->launch_description}; const std::array<u32, 3> dim{qmd.grid_dim_x, qmd.grid_dim_y, qmd.grid_dim_z}; scheduler.RequestOutsideRenderPassOperationContext(); scheduler.Record([dim](vk::CommandBuffer cmdbuf) { cmdbuf.Dispatch(dim[0], dim[1], dim[2]); }); @@ -422,7 +433,7 @@ void RasterizerVulkan::OnCPUWrite(VAddr addr, u64 size) { } } -void RasterizerVulkan::SyncGuestHost() { +void RasterizerVulkan::InvalidateGPUCache() { pipeline_cache.SyncGuestHost(); { std::scoped_lock lock{buffer_cache.mutex}; @@ -442,40 +453,30 @@ void RasterizerVulkan::UnmapMemory(VAddr addr, u64 size) { pipeline_cache.OnCPUWrite(addr, size); } -void RasterizerVulkan::ModifyGPUMemory(GPUVAddr addr, u64 size) { +void RasterizerVulkan::ModifyGPUMemory(size_t as_id, GPUVAddr addr, u64 size) { { std::scoped_lock lock{texture_cache.mutex}; - texture_cache.UnmapGPUMemory(addr, size); + texture_cache.UnmapGPUMemory(as_id, addr, size); } } -void RasterizerVulkan::SignalSemaphore(GPUVAddr addr, u32 value) { - if (!gpu.IsAsync()) { - gpu_memory.Write<u32>(addr, value); - return; - } - fence_manager.SignalSemaphore(addr, value); +void RasterizerVulkan::SignalFence(std::function<void()>&& func) { + fence_manager.SignalFence(std::move(func)); +} + +void RasterizerVulkan::SyncOperation(std::function<void()>&& func) { + fence_manager.SyncOperation(std::move(func)); } void RasterizerVulkan::SignalSyncPoint(u32 value) { - if (!gpu.IsAsync()) { - gpu.IncrementSyncPoint(value); - return; - } fence_manager.SignalSyncPoint(value); } void RasterizerVulkan::SignalReference() { - if (!gpu.IsAsync()) { - return; - } fence_manager.SignalOrdering(); } void RasterizerVulkan::ReleaseFences() { - if (!gpu.IsAsync()) { - return; - } fence_manager.WaitPendingFences(); } @@ -552,13 +553,13 @@ Tegra::Engines::AccelerateDMAInterface& RasterizerVulkan::AccessAccelerateDMA() } void RasterizerVulkan::AccelerateInlineToMemory(GPUVAddr address, size_t copy_size, - std::span<u8> memory) { - auto cpu_addr = gpu_memory.GpuToCpuAddress(address); + std::span<const u8> memory) { + auto cpu_addr = gpu_memory->GpuToCpuAddress(address); if (!cpu_addr) [[unlikely]] { - gpu_memory.WriteBlock(address, memory.data(), copy_size); + gpu_memory->WriteBlock(address, memory.data(), copy_size); return; } - gpu_memory.WriteBlockUnsafe(address, memory.data(), copy_size); + gpu_memory->WriteBlockUnsafe(address, memory.data(), copy_size); { std::unique_lock<std::mutex> lock{buffer_cache.mutex}; if (!buffer_cache.InlineMemory(*cpu_addr, copy_size, memory)) { @@ -627,7 +628,7 @@ bool AccelerateDMA::BufferCopy(GPUVAddr src_address, GPUVAddr dest_address, u64 } void RasterizerVulkan::UpdateDynamicStates() { - auto& regs = maxwell3d.regs; + auto& regs = maxwell3d->regs; UpdateViewportsState(regs); UpdateScissorsState(regs); UpdateDepthBias(regs); @@ -651,24 +652,24 @@ void RasterizerVulkan::UpdateDynamicStates() { } void RasterizerVulkan::BeginTransformFeedback() { - const auto& regs = maxwell3d.regs; - if (regs.tfb_enabled == 0) { + const auto& regs = maxwell3d->regs; + if (regs.transform_feedback_enabled == 0) { return; } if (!device.IsExtTransformFeedbackSupported()) { LOG_ERROR(Render_Vulkan, "Transform feedbacks used but not supported"); return; } - UNIMPLEMENTED_IF(regs.IsShaderConfigEnabled(Maxwell::ShaderProgram::TesselationControl) || - regs.IsShaderConfigEnabled(Maxwell::ShaderProgram::TesselationEval) || - regs.IsShaderConfigEnabled(Maxwell::ShaderProgram::Geometry)); + UNIMPLEMENTED_IF(regs.IsShaderConfigEnabled(Maxwell::ShaderType::TessellationInit) || + regs.IsShaderConfigEnabled(Maxwell::ShaderType::Tessellation) || + regs.IsShaderConfigEnabled(Maxwell::ShaderType::Geometry)); scheduler.Record( [](vk::CommandBuffer cmdbuf) { cmdbuf.BeginTransformFeedbackEXT(0, 0, nullptr, nullptr); }); } void RasterizerVulkan::EndTransformFeedback() { - const auto& regs = maxwell3d.regs; - if (regs.tfb_enabled == 0) { + const auto& regs = maxwell3d->regs; + if (regs.transform_feedback_enabled == 0) { return; } if (!device.IsExtTransformFeedbackSupported()) { @@ -748,11 +749,11 @@ void RasterizerVulkan::UpdateDepthBias(Tegra::Engines::Maxwell3D::Regs& regs) { if (!state_tracker.TouchDepthBias()) { return; } - float units = regs.polygon_offset_units / 2.0f; - const bool is_d24 = regs.zeta.format == Tegra::DepthFormat::S8_UINT_Z24_UNORM || - regs.zeta.format == Tegra::DepthFormat::D24X8_UNORM || - regs.zeta.format == Tegra::DepthFormat::D24S8_UNORM || - regs.zeta.format == Tegra::DepthFormat::D24C8_UNORM; + float units = regs.depth_bias / 2.0f; + const bool is_d24 = regs.zeta.format == Tegra::DepthFormat::Z24_UNORM_S8_UINT || + regs.zeta.format == Tegra::DepthFormat::X8Z24_UNORM || + regs.zeta.format == Tegra::DepthFormat::S8Z24_UNORM || + regs.zeta.format == Tegra::DepthFormat::V8Z24_UNORM; if (is_d24 && !device.SupportsD24DepthBuffer()) { // the base formulas can be obtained from here: // https://docs.microsoft.com/en-us/windows/win32/direct3d11/d3d10-graphics-programming-guide-output-merger-stage-depth-bias @@ -760,8 +761,8 @@ void RasterizerVulkan::UpdateDepthBias(Tegra::Engines::Maxwell3D::Regs& regs) { static_cast<double>(1ULL << (32 - 24)) / (static_cast<double>(0x1.ep+127)); units = static_cast<float>(static_cast<double>(units) * rescale_factor); } - scheduler.Record([constant = units, clamp = regs.polygon_offset_clamp, - factor = regs.polygon_offset_factor](vk::CommandBuffer cmdbuf) { + scheduler.Record([constant = units, clamp = regs.depth_bias_clamp, + factor = regs.slope_scale_depth_bias](vk::CommandBuffer cmdbuf) { cmdbuf.SetDepthBias(constant, clamp, factor); }); } @@ -791,8 +792,8 @@ void RasterizerVulkan::UpdateStencilFaces(Tegra::Engines::Maxwell3D::Regs& regs) if (regs.stencil_two_side_enable) { // Separate values per face scheduler.Record( - [front_ref = regs.stencil_front_func_ref, front_write_mask = regs.stencil_front_mask, - front_test_mask = regs.stencil_front_func_mask, back_ref = regs.stencil_back_func_ref, + [front_ref = regs.stencil_front_ref, front_write_mask = regs.stencil_front_mask, + front_test_mask = regs.stencil_front_func_mask, back_ref = regs.stencil_back_ref, back_write_mask = regs.stencil_back_mask, back_test_mask = regs.stencil_back_func_mask](vk::CommandBuffer cmdbuf) { // Front face @@ -807,7 +808,7 @@ void RasterizerVulkan::UpdateStencilFaces(Tegra::Engines::Maxwell3D::Regs& regs) }); } else { // Front face defines both faces - scheduler.Record([ref = regs.stencil_front_func_ref, write_mask = regs.stencil_front_mask, + scheduler.Record([ref = regs.stencil_front_ref, write_mask = regs.stencil_front_mask, test_mask = regs.stencil_front_func_mask](vk::CommandBuffer cmdbuf) { cmdbuf.SetStencilReference(VK_STENCIL_FACE_FRONT_AND_BACK, ref); cmdbuf.SetStencilWriteMask(VK_STENCIL_FACE_FRONT_AND_BACK, write_mask); @@ -820,7 +821,8 @@ void RasterizerVulkan::UpdateLineWidth(Tegra::Engines::Maxwell3D::Regs& regs) { if (!state_tracker.TouchLineWidth()) { return; } - const float width = regs.line_smooth_enable ? regs.line_width_smooth : regs.line_width_aliased; + const float width = + regs.line_anti_alias_enable ? regs.line_width_smooth : regs.line_width_aliased; scheduler.Record([width](vk::CommandBuffer cmdbuf) { cmdbuf.SetLineWidth(width); }); } @@ -828,10 +830,10 @@ void RasterizerVulkan::UpdateCullMode(Tegra::Engines::Maxwell3D::Regs& regs) { if (!state_tracker.TouchCullMode()) { return; } - scheduler.Record( - [enabled = regs.cull_test_enabled, cull_face = regs.cull_face](vk::CommandBuffer cmdbuf) { - cmdbuf.SetCullModeEXT(enabled ? MaxwellToVK::CullFace(cull_face) : VK_CULL_MODE_NONE); - }); + scheduler.Record([enabled = regs.gl_cull_test_enabled, + cull_face = regs.gl_cull_face](vk::CommandBuffer cmdbuf) { + cmdbuf.SetCullModeEXT(enabled ? MaxwellToVK::CullFace(cull_face) : VK_CULL_MODE_NONE); + }); } void RasterizerVulkan::UpdateDepthBoundsTestEnable(Tegra::Engines::Maxwell3D::Regs& regs) { @@ -880,8 +882,8 @@ void RasterizerVulkan::UpdateFrontFace(Tegra::Engines::Maxwell3D::Regs& regs) { return; } - VkFrontFace front_face = MaxwellToVK::FrontFace(regs.front_face); - if (regs.screen_y_control.triangle_rast_flip != 0) { + VkFrontFace front_face = MaxwellToVK::FrontFace(regs.gl_front_face); + if (regs.window_origin.flip_y != 0) { front_face = front_face == VK_FRONT_FACE_CLOCKWISE ? VK_FRONT_FACE_COUNTER_CLOCKWISE : VK_FRONT_FACE_CLOCKWISE; } @@ -893,16 +895,16 @@ void RasterizerVulkan::UpdateStencilOp(Tegra::Engines::Maxwell3D::Regs& regs) { if (!state_tracker.TouchStencilOp()) { return; } - const Maxwell::StencilOp fail = regs.stencil_front_op_fail; - const Maxwell::StencilOp zfail = regs.stencil_front_op_zfail; - const Maxwell::StencilOp zpass = regs.stencil_front_op_zpass; - const Maxwell::ComparisonOp compare = regs.stencil_front_func_func; + const Maxwell::StencilOp::Op fail = regs.stencil_front_op.fail; + const Maxwell::StencilOp::Op zfail = regs.stencil_front_op.zfail; + const Maxwell::StencilOp::Op zpass = regs.stencil_front_op.zpass; + const Maxwell::ComparisonOp compare = regs.stencil_front_op.func; if (regs.stencil_two_side_enable) { // Separate stencil op per face - const Maxwell::StencilOp back_fail = regs.stencil_back_op_fail; - const Maxwell::StencilOp back_zfail = regs.stencil_back_op_zfail; - const Maxwell::StencilOp back_zpass = regs.stencil_back_op_zpass; - const Maxwell::ComparisonOp back_compare = regs.stencil_back_func_func; + const Maxwell::StencilOp::Op back_fail = regs.stencil_back_op.fail; + const Maxwell::StencilOp::Op back_zfail = regs.stencil_back_op.zfail; + const Maxwell::StencilOp::Op back_zpass = regs.stencil_back_op.zpass; + const Maxwell::ComparisonOp back_compare = regs.stencil_back_op.func; scheduler.Record([fail, zfail, zpass, compare, back_fail, back_zfail, back_zpass, back_compare](vk::CommandBuffer cmdbuf) { cmdbuf.SetStencilOpEXT(VK_STENCIL_FACE_FRONT_BIT, MaxwellToVK::StencilOp(fail), @@ -933,7 +935,7 @@ void RasterizerVulkan::UpdateStencilTestEnable(Tegra::Engines::Maxwell3D::Regs& } void RasterizerVulkan::UpdateVertexInput(Tegra::Engines::Maxwell3D::Regs& regs) { - auto& dirty{maxwell3d.dirty.flags}; + auto& dirty{maxwell3d->dirty.flags}; if (!dirty[Dirty::VertexInput]) { return; } @@ -974,15 +976,15 @@ void RasterizerVulkan::UpdateVertexInput(Tegra::Engines::Maxwell3D::Regs& regs) dirty[Dirty::VertexBinding0 + index] = false; const u32 binding{static_cast<u32>(index)}; - const auto& input_binding{regs.vertex_array[binding]}; - const bool is_instanced{regs.instanced_arrays.IsInstancingEnabled(binding)}; + const auto& input_binding{regs.vertex_streams[binding]}; + const bool is_instanced{regs.vertex_stream_instances.IsInstancingEnabled(binding)}; bindings.push_back({ .sType = VK_STRUCTURE_TYPE_VERTEX_INPUT_BINDING_DESCRIPTION_2_EXT, .pNext = nullptr, .binding = binding, .stride = input_binding.stride, .inputRate = is_instanced ? VK_VERTEX_INPUT_RATE_INSTANCE : VK_VERTEX_INPUT_RATE_VERTEX, - .divisor = is_instanced ? input_binding.divisor : 1, + .divisor = is_instanced ? input_binding.frequency : 1, }); } scheduler.Record([bindings, attributes](vk::CommandBuffer cmdbuf) { @@ -990,4 +992,54 @@ void RasterizerVulkan::UpdateVertexInput(Tegra::Engines::Maxwell3D::Regs& regs) }); } +void RasterizerVulkan::InitializeChannel(Tegra::Control::ChannelState& channel) { + CreateChannel(channel); + { + std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex}; + texture_cache.CreateChannel(channel); + buffer_cache.CreateChannel(channel); + } + pipeline_cache.CreateChannel(channel); + query_cache.CreateChannel(channel); + state_tracker.SetupTables(channel); +} + +void RasterizerVulkan::BindChannel(Tegra::Control::ChannelState& channel) { + const s32 channel_id = channel.bind_id; + BindToChannel(channel_id); + { + std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex}; + texture_cache.BindToChannel(channel_id); + buffer_cache.BindToChannel(channel_id); + } + pipeline_cache.BindToChannel(channel_id); + query_cache.BindToChannel(channel_id); + state_tracker.ChangeChannel(channel); + state_tracker.InvalidateState(); +} + +void RasterizerVulkan::ReleaseChannel(s32 channel_id) { + EraseChannel(channel_id); + { + std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex}; + texture_cache.EraseChannel(channel_id); + buffer_cache.EraseChannel(channel_id); + } + pipeline_cache.EraseChannel(channel_id); + query_cache.EraseChannel(channel_id); +} + +void RasterizerVulkan::BindInlineIndexBuffer() { + if (maxwell3d->inline_index_draw_indexes.empty()) { + return; + } + const auto data_count = static_cast<u32>(maxwell3d->inline_index_draw_indexes.size()); + auto buffer = buffer_cache_runtime.UploadStagingBuffer(data_count); + std::memcpy(buffer.mapped_span.data(), maxwell3d->inline_index_draw_indexes.data(), data_count); + buffer_cache_runtime.BindIndexBuffer( + maxwell3d->regs.draw.topology, maxwell3d->regs.index_buffer.format, + maxwell3d->regs.index_buffer.first, maxwell3d->regs.index_buffer.count, buffer.buffer, + static_cast<u32>(buffer.offset), data_count); +} + } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.h b/src/video_core/renderer_vulkan/vk_rasterizer.h index 0370ea39b..b0bc306f5 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.h +++ b/src/video_core/renderer_vulkan/vk_rasterizer.h @@ -8,6 +8,7 @@ #include <boost/container/static_vector.hpp> #include "common/common_types.h" +#include "video_core/control/channel_state_cache.h" #include "video_core/engines/maxwell_dma.h" #include "video_core/rasterizer_accelerated.h" #include "video_core/rasterizer_interface.h" @@ -54,16 +55,16 @@ private: BufferCache& buffer_cache; }; -class RasterizerVulkan final : public VideoCore::RasterizerAccelerated { +class RasterizerVulkan final : public VideoCore::RasterizerAccelerated, + protected VideoCommon::ChannelSetupCaches<VideoCommon::ChannelInfo> { public: explicit RasterizerVulkan(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_, - Tegra::MemoryManager& gpu_memory_, Core::Memory::Memory& cpu_memory_, - ScreenInfo& screen_info_, const Device& device_, - MemoryAllocator& memory_allocator_, StateTracker& state_tracker_, - Scheduler& scheduler_); + Core::Memory::Memory& cpu_memory_, ScreenInfo& screen_info_, + const Device& device_, MemoryAllocator& memory_allocator_, + StateTracker& state_tracker_, Scheduler& scheduler_); ~RasterizerVulkan() override; - void Draw(bool is_indexed, bool is_instanced) override; + void Draw(bool is_indexed, u32 instance_count) override; void Clear() override; void DispatchCompute() override; void ResetCounter(VideoCore::QueryType type) override; @@ -75,10 +76,11 @@ public: bool MustFlushRegion(VAddr addr, u64 size) override; void InvalidateRegion(VAddr addr, u64 size) override; void OnCPUWrite(VAddr addr, u64 size) override; - void SyncGuestHost() override; + void InvalidateGPUCache() override; void UnmapMemory(VAddr addr, u64 size) override; - void ModifyGPUMemory(GPUVAddr addr, u64 size) override; - void SignalSemaphore(GPUVAddr addr, u32 value) override; + void ModifyGPUMemory(size_t as_id, GPUVAddr addr, u64 size) override; + void SignalFence(std::function<void()>&& func) override; + void SyncOperation(std::function<void()>&& func) override; void SignalSyncPoint(u32 value) override; void SignalReference() override; void ReleaseFences() override; @@ -93,12 +95,18 @@ public: const Tegra::Engines::Fermi2D::Config& copy_config) override; Tegra::Engines::AccelerateDMAInterface& AccessAccelerateDMA() override; void AccelerateInlineToMemory(GPUVAddr address, size_t copy_size, - std::span<u8> memory) override; + std::span<const u8> memory) override; bool AccelerateDisplay(const Tegra::FramebufferConfig& config, VAddr framebuffer_addr, u32 pixel_stride) override; void LoadDiskResources(u64 title_id, std::stop_token stop_loading, const VideoCore::DiskResourceLoadCallback& callback) override; + void InitializeChannel(Tegra::Control::ChannelState& channel) override; + + void BindChannel(Tegra::Control::ChannelState& channel) override; + + void ReleaseChannel(s32 channel_id) override; + private: static constexpr size_t MAX_TEXTURES = 192; static constexpr size_t MAX_IMAGES = 48; @@ -133,10 +141,9 @@ private: void UpdateVertexInput(Tegra::Engines::Maxwell3D::Regs& regs); + void BindInlineIndexBuffer(); + Tegra::GPU& gpu; - Tegra::MemoryManager& gpu_memory; - Tegra::Engines::Maxwell3D& maxwell3d; - Tegra::Engines::KeplerCompute& kepler_compute; ScreenInfo& screen_info; const Device& device; @@ -148,7 +155,6 @@ private: DescriptorPool descriptor_pool; UpdateDescriptorQueue update_descriptor_queue; BlitImageHelper blit_image; - ASTCDecoderPass astc_decoder_pass; RenderPassCache render_pass_cache; TextureCacheRuntime texture_cache_runtime; diff --git a/src/video_core/renderer_vulkan/vk_scheduler.cpp b/src/video_core/renderer_vulkan/vk_scheduler.cpp index a331ff37e..7934f2a51 100644 --- a/src/video_core/renderer_vulkan/vk_scheduler.cpp +++ b/src/video_core/renderer_vulkan/vk_scheduler.cpp @@ -136,9 +136,10 @@ bool Scheduler::UpdateRescaling(bool is_rescaling) { } void Scheduler::WorkerThread(std::stop_token stop_token) { - Common::SetCurrentThreadName("yuzu:VulkanWorker"); + Common::SetCurrentThreadName("VulkanWorker"); do { std::unique_ptr<CommandChunk> work; + bool has_submit{false}; { std::unique_lock lock{work_mutex}; if (work_queue.empty()) { @@ -150,9 +151,10 @@ void Scheduler::WorkerThread(std::stop_token stop_token) { } work = std::move(work_queue.front()); work_queue.pop(); + + has_submit = work->HasSubmit(); + work->ExecuteAll(current_cmdbuf); } - const bool has_submit = work->HasSubmit(); - work->ExecuteAll(current_cmdbuf); if (has_submit) { AllocateWorkerCommandBuffer(); } diff --git a/src/video_core/renderer_vulkan/vk_scheduler.h b/src/video_core/renderer_vulkan/vk_scheduler.h index c04aad08f..929216749 100644 --- a/src/video_core/renderer_vulkan/vk_scheduler.h +++ b/src/video_core/renderer_vulkan/vk_scheduler.h @@ -144,7 +144,6 @@ private: using FuncType = TypedCommand<T>; static_assert(sizeof(FuncType) < sizeof(data), "Lambda is too large"); - recorded_counts++; command_offset = Common::AlignUp(command_offset, alignof(FuncType)); if (command_offset > sizeof(data) - sizeof(FuncType)) { return false; @@ -166,7 +165,7 @@ private: } bool Empty() const { - return recorded_counts == 0; + return command_offset == 0; } bool HasSubmit() const { @@ -177,7 +176,6 @@ private: Command* first = nullptr; Command* last = nullptr; - size_t recorded_counts = 0; size_t command_offset = 0; bool submit = false; alignas(std::max_align_t) std::array<u8, 0x8000> data{}; diff --git a/src/video_core/renderer_vulkan/vk_state_tracker.cpp b/src/video_core/renderer_vulkan/vk_state_tracker.cpp index 9ad096431..b87c3be66 100644 --- a/src/video_core/renderer_vulkan/vk_state_tracker.cpp +++ b/src/video_core/renderer_vulkan/vk_state_tracker.cpp @@ -7,9 +7,9 @@ #include "common/common_types.h" #include "core/core.h" +#include "video_core/control/channel_state.h" #include "video_core/dirty_flags.h" #include "video_core/engines/maxwell_3d.h" -#include "video_core/gpu.h" #include "video_core/renderer_vulkan/vk_state_tracker.h" #define OFF(field_name) MAXWELL3D_REG_INDEX(field_name) @@ -51,8 +51,8 @@ Flags MakeInvalidationFlags() { void SetupDirtyViewports(Tables& tables) { FillBlock(tables[0], OFF(viewport_transform), NUM(viewport_transform), Viewports); FillBlock(tables[0], OFF(viewports), NUM(viewports), Viewports); - tables[0][OFF(viewport_transform_enabled)] = Viewports; - tables[1][OFF(screen_y_control)] = Viewports; + tables[0][OFF(viewport_scale_offset_enbled)] = Viewports; + tables[1][OFF(window_origin)] = Viewports; } void SetupDirtyScissors(Tables& tables) { @@ -61,9 +61,9 @@ void SetupDirtyScissors(Tables& tables) { void SetupDirtyDepthBias(Tables& tables) { auto& table = tables[0]; - table[OFF(polygon_offset_units)] = DepthBias; - table[OFF(polygon_offset_clamp)] = DepthBias; - table[OFF(polygon_offset_factor)] = DepthBias; + table[OFF(depth_bias)] = DepthBias; + table[OFF(depth_bias_clamp)] = DepthBias; + table[OFF(slope_scale_depth_bias)] = DepthBias; } void SetupDirtyBlendConstants(Tables& tables) { @@ -77,10 +77,10 @@ void SetupDirtyDepthBounds(Tables& tables) { void SetupDirtyStencilProperties(Tables& tables) { auto& table = tables[0]; table[OFF(stencil_two_side_enable)] = StencilProperties; - table[OFF(stencil_front_func_ref)] = StencilProperties; + table[OFF(stencil_front_ref)] = StencilProperties; table[OFF(stencil_front_mask)] = StencilProperties; table[OFF(stencil_front_func_mask)] = StencilProperties; - table[OFF(stencil_back_func_ref)] = StencilProperties; + table[OFF(stencil_back_ref)] = StencilProperties; table[OFF(stencil_back_mask)] = StencilProperties; table[OFF(stencil_back_func_mask)] = StencilProperties; } @@ -92,8 +92,8 @@ void SetupDirtyLineWidth(Tables& tables) { void SetupDirtyCullMode(Tables& tables) { auto& table = tables[0]; - table[OFF(cull_face)] = CullMode; - table[OFF(cull_test_enabled)] = CullMode; + table[OFF(gl_cull_face)] = CullMode; + table[OFF(gl_cull_test_enabled)] = CullMode; } void SetupDirtyDepthBoundsEnable(Tables& tables) { @@ -114,20 +114,20 @@ void SetupDirtyDepthCompareOp(Tables& tables) { void SetupDirtyFrontFace(Tables& tables) { auto& table = tables[0]; - table[OFF(front_face)] = FrontFace; - table[OFF(screen_y_control)] = FrontFace; + table[OFF(gl_front_face)] = FrontFace; + table[OFF(window_origin)] = FrontFace; } void SetupDirtyStencilOp(Tables& tables) { auto& table = tables[0]; - table[OFF(stencil_front_op_fail)] = StencilOp; - table[OFF(stencil_front_op_zfail)] = StencilOp; - table[OFF(stencil_front_op_zpass)] = StencilOp; - table[OFF(stencil_front_func_func)] = StencilOp; - table[OFF(stencil_back_op_fail)] = StencilOp; - table[OFF(stencil_back_op_zfail)] = StencilOp; - table[OFF(stencil_back_op_zpass)] = StencilOp; - table[OFF(stencil_back_func_func)] = StencilOp; + table[OFF(stencil_front_op.fail)] = StencilOp; + table[OFF(stencil_front_op.zfail)] = StencilOp; + table[OFF(stencil_front_op.zpass)] = StencilOp; + table[OFF(stencil_front_op.func)] = StencilOp; + table[OFF(stencil_back_op.fail)] = StencilOp; + table[OFF(stencil_back_op.zfail)] = StencilOp; + table[OFF(stencil_back_op.zpass)] = StencilOp; + table[OFF(stencil_back_op.func)] = StencilOp; // Table 0 is used by StencilProperties tables[1][OFF(stencil_two_side_enable)] = StencilOp; @@ -139,10 +139,10 @@ void SetupDirtyStencilTestEnable(Tables& tables) { void SetupDirtyBlending(Tables& tables) { tables[0][OFF(color_mask_common)] = Blending; - tables[0][OFF(independent_blend_enable)] = Blending; + tables[0][OFF(blend_per_target_enabled)] = Blending; FillBlock(tables[0], OFF(color_mask), NUM(color_mask), Blending); FillBlock(tables[0], OFF(blend), NUM(blend), Blending); - FillBlock(tables[0], OFF(independent_blend), NUM(independent_blend), Blending); + FillBlock(tables[0], OFF(blend_per_target), NUM(blend_per_target), Blending); } void SetupDirtyViewportSwizzles(Tables& tables) { @@ -166,17 +166,16 @@ void SetupDirtyVertexBindings(Tables& tables) { static constexpr size_t divisor_offset = 3; for (size_t i = 0; i < Regs::NumVertexArrays; ++i) { const u8 flag = static_cast<u8>(VertexBinding0 + i); - tables[0][OFF(instanced_arrays) + i] = VertexInput; - tables[1][OFF(instanced_arrays) + i] = flag; - tables[0][OFF(vertex_array) + i * NUM(vertex_array[0]) + divisor_offset] = VertexInput; - tables[1][OFF(vertex_array) + i * NUM(vertex_array[0]) + divisor_offset] = flag; + tables[0][OFF(vertex_stream_instances) + i] = VertexInput; + tables[1][OFF(vertex_stream_instances) + i] = flag; + tables[0][OFF(vertex_streams) + i * NUM(vertex_streams[0]) + divisor_offset] = VertexInput; + tables[1][OFF(vertex_streams) + i * NUM(vertex_streams[0]) + divisor_offset] = flag; } } } // Anonymous namespace -StateTracker::StateTracker(Tegra::GPU& gpu) - : flags{gpu.Maxwell3D().dirty.flags}, invalidation_flags{MakeInvalidationFlags()} { - auto& tables{gpu.Maxwell3D().dirty.tables}; +void StateTracker::SetupTables(Tegra::Control::ChannelState& channel_state) { + auto& tables{channel_state.maxwell_3d->dirty.tables}; SetupDirtyFlags(tables); SetupDirtyViewports(tables); SetupDirtyScissors(tables); @@ -199,4 +198,15 @@ StateTracker::StateTracker(Tegra::GPU& gpu) SetupDirtyVertexBindings(tables); } +void StateTracker::ChangeChannel(Tegra::Control::ChannelState& channel_state) { + flags = &channel_state.maxwell_3d->dirty.flags; +} + +void StateTracker::InvalidateState() { + flags->set(); +} + +StateTracker::StateTracker() + : flags{&default_flags}, default_flags{}, invalidation_flags{MakeInvalidationFlags()} {} + } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/vk_state_tracker.h b/src/video_core/renderer_vulkan/vk_state_tracker.h index a85bc1c10..2296dea60 100644 --- a/src/video_core/renderer_vulkan/vk_state_tracker.h +++ b/src/video_core/renderer_vulkan/vk_state_tracker.h @@ -10,6 +10,12 @@ #include "video_core/dirty_flags.h" #include "video_core/engines/maxwell_3d.h" +namespace Tegra { +namespace Control { +struct ChannelState; +} +} // namespace Tegra + namespace Vulkan { namespace Dirty { @@ -53,19 +59,19 @@ class StateTracker { using Maxwell = Tegra::Engines::Maxwell3D::Regs; public: - explicit StateTracker(Tegra::GPU& gpu); + explicit StateTracker(); void InvalidateCommandBufferState() { - flags |= invalidation_flags; + (*flags) |= invalidation_flags; current_topology = INVALID_TOPOLOGY; } void InvalidateViewports() { - flags[Dirty::Viewports] = true; + (*flags)[Dirty::Viewports] = true; } void InvalidateScissors() { - flags[Dirty::Scissors] = true; + (*flags)[Dirty::Scissors] = true; } bool TouchViewports() { @@ -139,16 +145,23 @@ public: return has_changed; } + void SetupTables(Tegra::Control::ChannelState& channel_state); + + void ChangeChannel(Tegra::Control::ChannelState& channel_state); + + void InvalidateState(); + private: static constexpr auto INVALID_TOPOLOGY = static_cast<Maxwell::PrimitiveTopology>(~0u); bool Exchange(std::size_t id, bool new_value) const noexcept { - const bool is_dirty = flags[id]; - flags[id] = new_value; + const bool is_dirty = (*flags)[id]; + (*flags)[id] = new_value; return is_dirty; } - Tegra::Engines::Maxwell3D::DirtyState::Flags& flags; + Tegra::Engines::Maxwell3D::DirtyState::Flags* flags; + Tegra::Engines::Maxwell3D::DirtyState::Flags default_flags; Tegra::Engines::Maxwell3D::DirtyState::Flags invalidation_flags; Maxwell::PrimitiveTopology current_topology = INVALID_TOPOLOGY; }; diff --git a/src/video_core/renderer_vulkan/vk_swapchain.cpp b/src/video_core/renderer_vulkan/vk_swapchain.cpp index a69ae7725..706d9ba74 100644 --- a/src/video_core/renderer_vulkan/vk_swapchain.cpp +++ b/src/video_core/renderer_vulkan/vk_swapchain.cpp @@ -36,7 +36,8 @@ VkPresentModeKHR ChooseSwapPresentMode(vk::Span<VkPresentModeKHR> modes) { // Mailbox (triple buffering) doesn't lock the application like fifo (vsync), // prefer it if vsync option is not selected const auto found_mailbox = std::find(modes.begin(), modes.end(), VK_PRESENT_MODE_MAILBOX_KHR); - if (found_mailbox != modes.end() && !Settings::values.use_vsync.GetValue()) { + if (Settings::values.fullscreen_mode.GetValue() == Settings::FullscreenMode::Borderless && + found_mailbox != modes.end() && !Settings::values.use_vsync.GetValue()) { return VK_PRESENT_MODE_MAILBOX_KHR; } if (!Settings::values.use_speed_limit.GetValue()) { @@ -156,8 +157,16 @@ void Swapchain::CreateSwapchain(const VkSurfaceCapabilitiesKHR& capabilities, u3 present_mode = ChooseSwapPresentMode(present_modes); u32 requested_image_count{capabilities.minImageCount + 1}; - if (capabilities.maxImageCount > 0 && requested_image_count > capabilities.maxImageCount) { - requested_image_count = capabilities.maxImageCount; + // Ensure Tripple buffering if possible. + if (capabilities.maxImageCount > 0) { + if (requested_image_count > capabilities.maxImageCount) { + requested_image_count = capabilities.maxImageCount; + } else { + requested_image_count = + std::max(requested_image_count, std::min(3U, capabilities.maxImageCount)); + } + } else { + requested_image_count = std::max(requested_image_count, 3U); } VkSwapchainCreateInfoKHR swapchain_ci{ .sType = VK_STRUCTURE_TYPE_SWAPCHAIN_CREATE_INFO_KHR, diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.cpp b/src/video_core/renderer_vulkan/vk_texture_cache.cpp index caca79d79..853b80d8a 100644 --- a/src/video_core/renderer_vulkan/vk_texture_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_texture_cache.cpp @@ -592,7 +592,7 @@ void TryTransformSwizzleIfNeeded(PixelFormat format, std::array<SwizzleSource, 4 case PixelFormat::A5B5G5R1_UNORM: std::ranges::transform(swizzle, swizzle.begin(), SwapSpecial); break; - case PixelFormat::R4G4_UNORM: + case PixelFormat::G4R4_UNORM: std::ranges::transform(swizzle, swizzle.begin(), SwapGreenRed); break; default: @@ -791,12 +791,17 @@ TextureCacheRuntime::TextureCacheRuntime(const Device& device_, Scheduler& sched MemoryAllocator& memory_allocator_, StagingBufferPool& staging_buffer_pool_, BlitImageHelper& blit_image_helper_, - ASTCDecoderPass& astc_decoder_pass_, - RenderPassCache& render_pass_cache_) + RenderPassCache& render_pass_cache_, + DescriptorPool& descriptor_pool, + UpdateDescriptorQueue& update_descriptor_queue) : device{device_}, scheduler{scheduler_}, memory_allocator{memory_allocator_}, staging_buffer_pool{staging_buffer_pool_}, blit_image_helper{blit_image_helper_}, - astc_decoder_pass{astc_decoder_pass_}, render_pass_cache{render_pass_cache_}, - resolution{Settings::values.resolution_info} {} + render_pass_cache{render_pass_cache_}, resolution{Settings::values.resolution_info} { + if (Settings::values.accelerate_astc) { + astc_decoder_pass.emplace(device, scheduler, descriptor_pool, staging_buffer_pool, + update_descriptor_queue, memory_allocator); + } +} void TextureCacheRuntime::Finish() { scheduler.Finish(); @@ -1474,13 +1479,14 @@ bool Image::BlitScaleHelper(bool scale_up) { }; const VkExtent2D extent{ .width = std::max(scaled_width, info.size.width), - .height = std::max(scaled_height, info.size.width), + .height = std::max(scaled_height, info.size.height), }; auto* view_ptr = blit_view.get(); if (aspect_mask == VK_IMAGE_ASPECT_COLOR_BIT) { if (!blit_framebuffer) { - blit_framebuffer = std::make_unique<Framebuffer>(*runtime, view_ptr, nullptr, extent); + blit_framebuffer = + std::make_unique<Framebuffer>(*runtime, view_ptr, nullptr, extent, scale_up); } const auto color_view = blit_view->Handle(Shader::TextureType::Color2D); @@ -1488,7 +1494,8 @@ bool Image::BlitScaleHelper(bool scale_up) { src_region, operation, BLIT_OPERATION); } else if (aspect_mask == (VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT)) { if (!blit_framebuffer) { - blit_framebuffer = std::make_unique<Framebuffer>(*runtime, nullptr, view_ptr, extent); + blit_framebuffer = + std::make_unique<Framebuffer>(*runtime, nullptr, view_ptr, extent, scale_up); } runtime->blit_image_helper.BlitDepthStencil(blit_framebuffer.get(), blit_view->DepthView(), blit_view->StencilView(), dst_region, @@ -1756,34 +1763,42 @@ Framebuffer::Framebuffer(TextureCacheRuntime& runtime, std::span<ImageView*, NUM .width = key.size.width, .height = key.size.height, }} { - CreateFramebuffer(runtime, color_buffers, depth_buffer); + CreateFramebuffer(runtime, color_buffers, depth_buffer, key.is_rescaled); if (runtime.device.HasDebuggingToolAttached()) { framebuffer.SetObjectNameEXT(VideoCommon::Name(key).c_str()); } } Framebuffer::Framebuffer(TextureCacheRuntime& runtime, ImageView* color_buffer, - ImageView* depth_buffer, VkExtent2D extent) + ImageView* depth_buffer, VkExtent2D extent, bool is_rescaled) : render_area{extent} { std::array<ImageView*, NUM_RT> color_buffers{color_buffer}; - CreateFramebuffer(runtime, color_buffers, depth_buffer); + CreateFramebuffer(runtime, color_buffers, depth_buffer, is_rescaled); } Framebuffer::~Framebuffer() = default; void Framebuffer::CreateFramebuffer(TextureCacheRuntime& runtime, std::span<ImageView*, NUM_RT> color_buffers, - ImageView* depth_buffer) { + ImageView* depth_buffer, bool is_rescaled) { std::vector<VkImageView> attachments; RenderPassKey renderpass_key{}; s32 num_layers = 1; + const auto& resolution = runtime.resolution; + + u32 width = std::numeric_limits<u32>::max(); + u32 height = std::numeric_limits<u32>::max(); for (size_t index = 0; index < NUM_RT; ++index) { const ImageView* const color_buffer = color_buffers[index]; if (!color_buffer) { renderpass_key.color_formats[index] = PixelFormat::Invalid; continue; } + width = std::min(width, is_rescaled ? resolution.ScaleUp(color_buffer->size.width) + : color_buffer->size.width); + height = std::min(height, is_rescaled ? resolution.ScaleUp(color_buffer->size.height) + : color_buffer->size.height); attachments.push_back(color_buffer->RenderTarget()); renderpass_key.color_formats[index] = color_buffer->format; num_layers = std::max(num_layers, color_buffer->range.extent.layers); @@ -1794,6 +1809,10 @@ void Framebuffer::CreateFramebuffer(TextureCacheRuntime& runtime, } const size_t num_colors = attachments.size(); if (depth_buffer) { + width = std::min(width, is_rescaled ? resolution.ScaleUp(depth_buffer->size.width) + : depth_buffer->size.width); + height = std::min(height, is_rescaled ? resolution.ScaleUp(depth_buffer->size.height) + : depth_buffer->size.height); attachments.push_back(depth_buffer->RenderTarget()); renderpass_key.depth_format = depth_buffer->format; num_layers = std::max(num_layers, depth_buffer->range.extent.layers); @@ -1810,6 +1829,8 @@ void Framebuffer::CreateFramebuffer(TextureCacheRuntime& runtime, renderpass_key.samples = samples; renderpass = runtime.render_pass_cache.Get(renderpass_key); + render_area.width = std::min(render_area.width, width); + render_area.height = std::min(render_area.height, height); num_color_buffers = static_cast<u32>(num_colors); framebuffer = runtime.device.GetLogical().CreateFramebuffer({ @@ -1829,7 +1850,7 @@ void TextureCacheRuntime::AccelerateImageUpload( Image& image, const StagingBufferRef& map, std::span<const VideoCommon::SwizzleParameters> swizzles) { if (IsPixelFormatASTC(image.info.format)) { - return astc_decoder_pass.Assemble(image, map, swizzles); + return astc_decoder_pass->Assemble(image, map, swizzles); } ASSERT(false); } diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.h b/src/video_core/renderer_vulkan/vk_texture_cache.h index 69f06ee7b..7ec0df134 100644 --- a/src/video_core/renderer_vulkan/vk_texture_cache.h +++ b/src/video_core/renderer_vulkan/vk_texture_cache.h @@ -6,6 +6,7 @@ #include <span> #include "shader_recompiler/shader_info.h" +#include "video_core/renderer_vulkan/vk_compute_pass.h" #include "video_core/renderer_vulkan/vk_staging_buffer_pool.h" #include "video_core/texture_cache/image_view_base.h" #include "video_core/texture_cache/texture_cache_base.h" @@ -25,14 +26,15 @@ using VideoCommon::RenderTargets; using VideoCommon::SlotVector; using VideoCore::Surface::PixelFormat; -class ASTCDecoderPass; class BlitImageHelper; +class DescriptorPool; class Device; class Image; class ImageView; class Framebuffer; class RenderPassCache; class StagingBufferPool; +class UpdateDescriptorQueue; class Scheduler; class TextureCacheRuntime { @@ -41,8 +43,9 @@ public: MemoryAllocator& memory_allocator_, StagingBufferPool& staging_buffer_pool_, BlitImageHelper& blit_image_helper_, - ASTCDecoderPass& astc_decoder_pass_, - RenderPassCache& render_pass_cache_); + RenderPassCache& render_pass_cache_, + DescriptorPool& descriptor_pool, + UpdateDescriptorQueue& update_descriptor_queue); void Finish(); @@ -97,8 +100,8 @@ public: MemoryAllocator& memory_allocator; StagingBufferPool& staging_buffer_pool; BlitImageHelper& blit_image_helper; - ASTCDecoderPass& astc_decoder_pass; RenderPassCache& render_pass_cache; + std::optional<ASTCDecoderPass> astc_decoder_pass; const Settings::ResolutionScalingInfo& resolution; constexpr static size_t indexing_slots = 8 * sizeof(size_t); @@ -268,7 +271,7 @@ public: ImageView* depth_buffer, const VideoCommon::RenderTargets& key); explicit Framebuffer(TextureCacheRuntime& runtime, ImageView* color_buffer, - ImageView* depth_buffer, VkExtent2D extent); + ImageView* depth_buffer, VkExtent2D extent, bool is_rescaled); ~Framebuffer(); @@ -279,7 +282,8 @@ public: Framebuffer& operator=(Framebuffer&&) = default; void CreateFramebuffer(TextureCacheRuntime& runtime, - std::span<ImageView*, NUM_RT> color_buffers, ImageView* depth_buffer); + std::span<ImageView*, NUM_RT> color_buffers, ImageView* depth_buffer, + bool is_rescaled = false); [[nodiscard]] VkFramebuffer Handle() const noexcept { return *framebuffer; |