diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/video_core/engines/maxwell_3d.cpp | 66 | ||||
| -rw-r--r-- | src/video_core/engines/maxwell_3d.h | 15 | ||||
| -rw-r--r-- | src/video_core/renderer_opengl/gl_rasterizer.cpp | 18 | ||||
| -rw-r--r-- | src/video_core/renderer_opengl/gl_rasterizer_cache.cpp | 4 |
4 files changed, 51 insertions, 52 deletions
diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp index 2d2136067..144e7fa82 100644 --- a/src/video_core/engines/maxwell_3d.cpp +++ b/src/video_core/engines/maxwell_3d.cpp | |||
| @@ -107,21 +107,23 @@ void Maxwell3D::CallMacroMethod(u32 method, std::vector<u32> parameters) { | |||
| 107 | void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) { | 107 | void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) { |
| 108 | auto debug_context = system.GetGPUDebugContext(); | 108 | auto debug_context = system.GetGPUDebugContext(); |
| 109 | 109 | ||
| 110 | const u32 method = method_call.method; | ||
| 111 | |||
| 110 | // It is an error to write to a register other than the current macro's ARG register before it | 112 | // It is an error to write to a register other than the current macro's ARG register before it |
| 111 | // has finished execution. | 113 | // has finished execution. |
| 112 | if (executing_macro != 0) { | 114 | if (executing_macro != 0) { |
| 113 | ASSERT(method_call.method == executing_macro + 1); | 115 | ASSERT(method == executing_macro + 1); |
| 114 | } | 116 | } |
| 115 | 117 | ||
| 116 | // Methods after 0xE00 are special, they're actually triggers for some microcode that was | 118 | // Methods after 0xE00 are special, they're actually triggers for some microcode that was |
| 117 | // uploaded to the GPU during initialization. | 119 | // uploaded to the GPU during initialization. |
| 118 | if (method_call.method >= MacroRegistersStart) { | 120 | if (method >= MacroRegistersStart) { |
| 119 | // We're trying to execute a macro | 121 | // We're trying to execute a macro |
| 120 | if (executing_macro == 0) { | 122 | if (executing_macro == 0) { |
| 121 | // A macro call must begin by writing the macro method's register, not its argument. | 123 | // A macro call must begin by writing the macro method's register, not its argument. |
| 122 | ASSERT_MSG((method_call.method % 2) == 0, | 124 | ASSERT_MSG((method % 2) == 0, |
| 123 | "Can't start macro execution by writing to the ARGS register"); | 125 | "Can't start macro execution by writing to the ARGS register"); |
| 124 | executing_macro = method_call.method; | 126 | executing_macro = method; |
| 125 | } | 127 | } |
| 126 | 128 | ||
| 127 | macro_params.push_back(method_call.argument); | 129 | macro_params.push_back(method_call.argument); |
| @@ -133,66 +135,62 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) { | |||
| 133 | return; | 135 | return; |
| 134 | } | 136 | } |
| 135 | 137 | ||
| 136 | ASSERT_MSG(method_call.method < Regs::NUM_REGS, | 138 | ASSERT_MSG(method < Regs::NUM_REGS, |
| 137 | "Invalid Maxwell3D register, increase the size of the Regs structure"); | 139 | "Invalid Maxwell3D register, increase the size of the Regs structure"); |
| 138 | 140 | ||
| 139 | if (debug_context) { | 141 | if (debug_context) { |
| 140 | debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandLoaded, nullptr); | 142 | debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandLoaded, nullptr); |
| 141 | } | 143 | } |
| 142 | 144 | ||
| 143 | if (regs.reg_array[method_call.method] != method_call.argument) { | 145 | if (regs.reg_array[method] != method_call.argument) { |
| 144 | regs.reg_array[method_call.method] = method_call.argument; | 146 | regs.reg_array[method] = method_call.argument; |
| 145 | // Color buffers | 147 | // Color buffers |
| 146 | constexpr u32 first_rt_reg = MAXWELL3D_REG_INDEX(rt); | 148 | constexpr u32 first_rt_reg = MAXWELL3D_REG_INDEX(rt); |
| 147 | constexpr u32 registers_per_rt = sizeof(regs.rt[0]) / sizeof(u32); | 149 | constexpr u32 registers_per_rt = sizeof(regs.rt[0]) / sizeof(u32); |
| 148 | if (method_call.method >= first_rt_reg && | 150 | if (method >= first_rt_reg && |
| 149 | method_call.method < first_rt_reg + registers_per_rt * Regs::NumRenderTargets) { | 151 | method < first_rt_reg + registers_per_rt * Regs::NumRenderTargets) { |
| 150 | const std::size_t rt_index = (method_call.method - first_rt_reg) / registers_per_rt; | 152 | const std::size_t rt_index = (method - first_rt_reg) / registers_per_rt; |
| 151 | dirty_flags.color_buffer |= 1u << static_cast<u32>(rt_index); | 153 | dirty_flags.color_buffer.set(rt_index); |
| 152 | } | 154 | } |
| 153 | 155 | ||
| 154 | // Zeta buffer | 156 | // Zeta buffer |
| 155 | constexpr u32 registers_in_zeta = sizeof(regs.zeta) / sizeof(u32); | 157 | constexpr u32 registers_in_zeta = sizeof(regs.zeta) / sizeof(u32); |
| 156 | if (method_call.method == MAXWELL3D_REG_INDEX(zeta_enable) || | 158 | if (method == MAXWELL3D_REG_INDEX(zeta_enable) || |
| 157 | method_call.method == MAXWELL3D_REG_INDEX(zeta_width) || | 159 | method == MAXWELL3D_REG_INDEX(zeta_width) || |
| 158 | method_call.method == MAXWELL3D_REG_INDEX(zeta_height) || | 160 | method == MAXWELL3D_REG_INDEX(zeta_height) || |
| 159 | (method_call.method >= MAXWELL3D_REG_INDEX(zeta) && | 161 | (method >= MAXWELL3D_REG_INDEX(zeta) && |
| 160 | method_call.method < MAXWELL3D_REG_INDEX(zeta) + registers_in_zeta)) { | 162 | method < MAXWELL3D_REG_INDEX(zeta) + registers_in_zeta)) { |
| 161 | dirty_flags.zeta_buffer = true; | 163 | dirty_flags.zeta_buffer = true; |
| 162 | } | 164 | } |
| 163 | 165 | ||
| 164 | // Shader | 166 | // Shader |
| 165 | constexpr u32 shader_registers_count = | 167 | constexpr u32 shader_registers_count = |
| 166 | sizeof(regs.shader_config[0]) * Regs::MaxShaderProgram / sizeof(u32); | 168 | sizeof(regs.shader_config[0]) * Regs::MaxShaderProgram / sizeof(u32); |
| 167 | if (method_call.method >= MAXWELL3D_REG_INDEX(shader_config[0]) && | 169 | if (method >= MAXWELL3D_REG_INDEX(shader_config[0]) && |
| 168 | method_call.method < MAXWELL3D_REG_INDEX(shader_config[0]) + shader_registers_count) { | 170 | method < MAXWELL3D_REG_INDEX(shader_config[0]) + shader_registers_count) { |
| 169 | dirty_flags.shaders = true; | 171 | dirty_flags.shaders = true; |
| 170 | } | 172 | } |
| 171 | 173 | ||
| 172 | // Vertex format | 174 | // Vertex format |
| 173 | if (method_call.method >= MAXWELL3D_REG_INDEX(vertex_attrib_format) && | 175 | if (method >= MAXWELL3D_REG_INDEX(vertex_attrib_format) && |
| 174 | method_call.method < | 176 | method < MAXWELL3D_REG_INDEX(vertex_attrib_format) + regs.vertex_attrib_format.size()) { |
| 175 | MAXWELL3D_REG_INDEX(vertex_attrib_format) + regs.vertex_attrib_format.size()) { | ||
| 176 | dirty_flags.vertex_attrib_format = true; | 177 | dirty_flags.vertex_attrib_format = true; |
| 177 | } | 178 | } |
| 178 | 179 | ||
| 179 | // Vertex buffer | 180 | // Vertex buffer |
| 180 | if (method_call.method >= MAXWELL3D_REG_INDEX(vertex_array) && | 181 | if (method >= MAXWELL3D_REG_INDEX(vertex_array) && |
| 181 | method_call.method < MAXWELL3D_REG_INDEX(vertex_array) + 4 * 32) { | 182 | method < MAXWELL3D_REG_INDEX(vertex_array) + 4 * 32) { |
| 182 | dirty_flags.vertex_array |= | 183 | dirty_flags.vertex_array.set((method - MAXWELL3D_REG_INDEX(vertex_array)) >> 2); |
| 183 | 1u << ((method_call.method - MAXWELL3D_REG_INDEX(vertex_array)) >> 2); | 184 | } else if (method >= MAXWELL3D_REG_INDEX(vertex_array_limit) && |
| 184 | } else if (method_call.method >= MAXWELL3D_REG_INDEX(vertex_array_limit) && | 185 | method < MAXWELL3D_REG_INDEX(vertex_array_limit) + 2 * 32) { |
| 185 | method_call.method < MAXWELL3D_REG_INDEX(vertex_array_limit) + 2 * 32) { | 186 | dirty_flags.vertex_array.set((method - MAXWELL3D_REG_INDEX(vertex_array_limit)) >> 1); |
| 186 | dirty_flags.vertex_array |= | 187 | } else if (method >= MAXWELL3D_REG_INDEX(instanced_arrays) && |
| 187 | 1u << ((method_call.method - MAXWELL3D_REG_INDEX(vertex_array_limit)) >> 1); | 188 | method < MAXWELL3D_REG_INDEX(instanced_arrays) + 32) { |
| 188 | } else if (method_call.method >= MAXWELL3D_REG_INDEX(instanced_arrays) && | 189 | dirty_flags.vertex_array.set(method - MAXWELL3D_REG_INDEX(instanced_arrays)); |
| 189 | method_call.method < MAXWELL3D_REG_INDEX(instanced_arrays) + 32) { | ||
| 190 | dirty_flags.vertex_array |= | ||
| 191 | 1u << (method_call.method - MAXWELL3D_REG_INDEX(instanced_arrays)); | ||
| 192 | } | 190 | } |
| 193 | } | 191 | } |
| 194 | 192 | ||
| 195 | switch (method_call.method) { | 193 | switch (method) { |
| 196 | case MAXWELL3D_REG_INDEX(macros.data): { | 194 | case MAXWELL3D_REG_INDEX(macros.data): { |
| 197 | ProcessMacroUpload(method_call.argument); | 195 | ProcessMacroUpload(method_call.argument); |
| 198 | break; | 196 | break; |
diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h index 584f51c48..7fbf1026e 100644 --- a/src/video_core/engines/maxwell_3d.h +++ b/src/video_core/engines/maxwell_3d.h | |||
| @@ -5,8 +5,10 @@ | |||
| 5 | #pragma once | 5 | #pragma once |
| 6 | 6 | ||
| 7 | #include <array> | 7 | #include <array> |
| 8 | #include <bitset> | ||
| 8 | #include <unordered_map> | 9 | #include <unordered_map> |
| 9 | #include <vector> | 10 | #include <vector> |
| 11 | |||
| 10 | #include "common/assert.h" | 12 | #include "common/assert.h" |
| 11 | #include "common/bit_field.h" | 13 | #include "common/bit_field.h" |
| 12 | #include "common/common_funcs.h" | 14 | #include "common/common_funcs.h" |
| @@ -1094,19 +1096,18 @@ public: | |||
| 1094 | MemoryManager& memory_manager; | 1096 | MemoryManager& memory_manager; |
| 1095 | 1097 | ||
| 1096 | struct DirtyFlags { | 1098 | struct DirtyFlags { |
| 1097 | u8 color_buffer = 0xFF; | 1099 | std::bitset<8> color_buffer{0xFF}; |
| 1098 | bool zeta_buffer = true; | 1100 | std::bitset<32> vertex_array{0xFFFFFFFF}; |
| 1099 | |||
| 1100 | bool shaders = true; | ||
| 1101 | 1101 | ||
| 1102 | bool vertex_attrib_format = true; | 1102 | bool vertex_attrib_format = true; |
| 1103 | u32 vertex_array = 0xFFFFFFFF; | 1103 | bool zeta_buffer = true; |
| 1104 | bool shaders = true; | ||
| 1104 | 1105 | ||
| 1105 | void OnMemoryWrite() { | 1106 | void OnMemoryWrite() { |
| 1106 | color_buffer = 0xFF; | ||
| 1107 | zeta_buffer = true; | 1107 | zeta_buffer = true; |
| 1108 | shaders = true; | 1108 | shaders = true; |
| 1109 | vertex_array = 0xFFFFFFFF; | 1109 | color_buffer.set(); |
| 1110 | vertex_array.set(); | ||
| 1110 | } | 1111 | } |
| 1111 | }; | 1112 | }; |
| 1112 | 1113 | ||
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 09fa01d25..ee04569fa 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp | |||
| @@ -200,7 +200,7 @@ GLuint RasterizerOpenGL::SetupVertexFormat() { | |||
| 200 | } | 200 | } |
| 201 | 201 | ||
| 202 | // Rebinding the VAO invalidates the vertex buffer bindings. | 202 | // Rebinding the VAO invalidates the vertex buffer bindings. |
| 203 | gpu.dirty_flags.vertex_array = 0xFFFFFFFF; | 203 | gpu.dirty_flags.vertex_array.set(); |
| 204 | 204 | ||
| 205 | state.draw.vertex_array = vao_entry.handle; | 205 | state.draw.vertex_array = vao_entry.handle; |
| 206 | return vao_entry.handle; | 206 | return vao_entry.handle; |
| @@ -210,14 +210,14 @@ void RasterizerOpenGL::SetupVertexBuffer(GLuint vao) { | |||
| 210 | auto& gpu = Core::System::GetInstance().GPU().Maxwell3D(); | 210 | auto& gpu = Core::System::GetInstance().GPU().Maxwell3D(); |
| 211 | const auto& regs = gpu.regs; | 211 | const auto& regs = gpu.regs; |
| 212 | 212 | ||
| 213 | if (!gpu.dirty_flags.vertex_array) | 213 | if (gpu.dirty_flags.vertex_array.none()) |
| 214 | return; | 214 | return; |
| 215 | 215 | ||
| 216 | MICROPROFILE_SCOPE(OpenGL_VB); | 216 | MICROPROFILE_SCOPE(OpenGL_VB); |
| 217 | 217 | ||
| 218 | // Upload all guest vertex arrays sequentially to our buffer | 218 | // Upload all guest vertex arrays sequentially to our buffer |
| 219 | for (u32 index = 0; index < Maxwell::NumVertexArrays; ++index) { | 219 | for (u32 index = 0; index < Maxwell::NumVertexArrays; ++index) { |
| 220 | if (~gpu.dirty_flags.vertex_array & (1u << index)) | 220 | if (!gpu.dirty_flags.vertex_array[index]) |
| 221 | continue; | 221 | continue; |
| 222 | 222 | ||
| 223 | const auto& vertex_array = regs.vertex_array[index]; | 223 | const auto& vertex_array = regs.vertex_array[index]; |
| @@ -244,7 +244,7 @@ void RasterizerOpenGL::SetupVertexBuffer(GLuint vao) { | |||
| 244 | } | 244 | } |
| 245 | } | 245 | } |
| 246 | 246 | ||
| 247 | gpu.dirty_flags.vertex_array = 0; | 247 | gpu.dirty_flags.vertex_array.reset(); |
| 248 | } | 248 | } |
| 249 | 249 | ||
| 250 | DrawParameters RasterizerOpenGL::SetupDraw() { | 250 | DrawParameters RasterizerOpenGL::SetupDraw() { |
| @@ -488,13 +488,13 @@ std::pair<bool, bool> RasterizerOpenGL::ConfigureFramebuffers( | |||
| 488 | OpenGLState& current_state, bool using_color_fb, bool using_depth_fb, bool preserve_contents, | 488 | OpenGLState& current_state, bool using_color_fb, bool using_depth_fb, bool preserve_contents, |
| 489 | std::optional<std::size_t> single_color_target) { | 489 | std::optional<std::size_t> single_color_target) { |
| 490 | MICROPROFILE_SCOPE(OpenGL_Framebuffer); | 490 | MICROPROFILE_SCOPE(OpenGL_Framebuffer); |
| 491 | const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D(); | 491 | auto& gpu = Core::System::GetInstance().GPU().Maxwell3D(); |
| 492 | const auto& regs = gpu.regs; | 492 | const auto& regs = gpu.regs; |
| 493 | 493 | ||
| 494 | const FramebufferConfigState fb_config_state{using_color_fb, using_depth_fb, preserve_contents, | 494 | const FramebufferConfigState fb_config_state{using_color_fb, using_depth_fb, preserve_contents, |
| 495 | single_color_target}; | 495 | single_color_target}; |
| 496 | if (fb_config_state == current_framebuffer_config_state && gpu.dirty_flags.color_buffer == 0 && | 496 | if (fb_config_state == current_framebuffer_config_state && |
| 497 | !gpu.dirty_flags.zeta_buffer) { | 497 | gpu.dirty_flags.color_buffer.none() && !gpu.dirty_flags.zeta_buffer) { |
| 498 | // Only skip if the previous ConfigureFramebuffers call was from the same kind (multiple or | 498 | // Only skip if the previous ConfigureFramebuffers call was from the same kind (multiple or |
| 499 | // single color targets). This is done because the guest registers may not change but the | 499 | // single color targets). This is done because the guest registers may not change but the |
| 500 | // host framebuffer may contain different attachments | 500 | // host framebuffer may contain different attachments |
| @@ -721,10 +721,10 @@ void RasterizerOpenGL::DrawArrays() { | |||
| 721 | // Add space for at least 18 constant buffers | 721 | // Add space for at least 18 constant buffers |
| 722 | buffer_size += Maxwell::MaxConstBuffers * (MaxConstbufferSize + uniform_buffer_alignment); | 722 | buffer_size += Maxwell::MaxConstBuffers * (MaxConstbufferSize + uniform_buffer_alignment); |
| 723 | 723 | ||
| 724 | bool invalidate = buffer_cache.Map(buffer_size); | 724 | const bool invalidate = buffer_cache.Map(buffer_size); |
| 725 | if (invalidate) { | 725 | if (invalidate) { |
| 726 | // As all cached buffers are invalidated, we need to recheck their state. | 726 | // As all cached buffers are invalidated, we need to recheck their state. |
| 727 | gpu.dirty_flags.vertex_array = 0xFFFFFFFF; | 727 | gpu.dirty_flags.vertex_array.set(); |
| 728 | } | 728 | } |
| 729 | 729 | ||
| 730 | const GLuint vao = SetupVertexFormat(); | 730 | const GLuint vao = SetupVertexFormat(); |
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp index 49c79811d..81b6099f9 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp | |||
| @@ -962,10 +962,10 @@ Surface RasterizerCacheOpenGL::GetColorBufferSurface(std::size_t index, bool pre | |||
| 962 | auto& gpu{Core::System::GetInstance().GPU().Maxwell3D()}; | 962 | auto& gpu{Core::System::GetInstance().GPU().Maxwell3D()}; |
| 963 | const auto& regs{gpu.regs}; | 963 | const auto& regs{gpu.regs}; |
| 964 | 964 | ||
| 965 | if ((gpu.dirty_flags.color_buffer & (1u << static_cast<u32>(index))) == 0) { | 965 | if (!gpu.dirty_flags.color_buffer[index]) { |
| 966 | return last_color_buffers[index]; | 966 | return last_color_buffers[index]; |
| 967 | } | 967 | } |
| 968 | gpu.dirty_flags.color_buffer &= ~(1u << static_cast<u32>(index)); | 968 | gpu.dirty_flags.color_buffer.reset(index); |
| 969 | 969 | ||
| 970 | ASSERT(index < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets); | 970 | ASSERT(index < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets); |
| 971 | 971 | ||