early-access version 4007

README.md

@@ -1,7 +1,7 @@
 yuzu emulator early access
 =============
 
-This is the source code for early-access 4006.
+This is the source code for early-access 4007.
 
 ## Legal Notice
 

src/shader_recompiler/backend/glasm/emit_glasm_context_get_set.cpp

@@ -7,7 +7,6 @@
 #include "shader_recompiler/backend/glasm/glasm_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
 #include "shader_recompiler/profile.h"
-#include "shader_recompiler/runtime_info.h"
 #include "shader_recompiler/shader_info.h"
 
 namespace Shader::Backend::GLASM {
@@ -24,14 +23,7 @@ void GetCbuf(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding, ScalarU
     }
 
     if (binding.IsImmediate()) {
-        const u32 binding_index{binding.U32()};
+        ctx.Add("LDC.{} {},c{}[{}];", size, ret, binding.U32(), offset);
-        const u32 max_num_cbufs{ctx.runtime_info.max_num_cbufs};
-        if (binding_index >= max_num_cbufs) {
-            // cbuf index exceeds device limit
-            ctx.Add("MOV.S {},0;", ret);
-            return;
-        }
-        ctx.Add("LDC.{} {},c{}[{}];", size, ret, binding_index, offset);
         return;
     }
 

src/shader_recompiler/backend/glasm/glasm_emit_context.cpp

@@ -37,12 +37,6 @@ EmitContext::EmitContext(IR::Program& program, Bindings& bindings, const Profile
         if (desc.count != 1) {
             throw NotImplementedException("Constant buffer descriptor array");
         }
-        if (cbuf_index >= runtime_info.max_num_cbufs) {
-            LOG_WARNING(Shader_GLASM, "Constant buffer binding index {} exceeds device limit of {}",
-                        cbuf_index, runtime_info.max_num_cbufs);
-            ++cbuf_index;
-            continue;
-        }
         Add("CBUFFER c{}[]={{program.buffer[{}]}};", desc.index, cbuf_index);
         ++cbuf_index;
     }

src/shader_recompiler/backend/glsl/emit_glsl.h

@@ -15,10 +15,9 @@ namespace Shader::Backend::GLSL {
 [[nodiscard]] std::string EmitGLSL(const Profile& profile, const RuntimeInfo& runtime_info,
                                    IR::Program& program, Bindings& bindings);
 
-[[nodiscard]] inline std::string EmitGLSL(const Profile& profile, const RuntimeInfo& runtime_info,
+[[nodiscard]] inline std::string EmitGLSL(const Profile& profile, IR::Program& program) {
-                                          IR::Program& program) {
     Bindings binding;
-    return EmitGLSL(profile, runtime_info, program, binding);
+    return EmitGLSL(profile, {}, program, binding);
 }
 
 } // namespace Shader::Backend::GLSL

src/shader_recompiler/backend/glsl/emit_glsl_context_get_set.cpp

@@ -46,15 +46,6 @@ std::string ChooseCbuf(EmitContext& ctx, const IR::Value& binding, std::string_v
 void GetCbuf(EmitContext& ctx, std::string_view ret, const IR::Value& binding,
              const IR::Value& offset, u32 num_bits, std::string_view cast = {},
              std::string_view bit_offset = {}) {
-    if (binding.IsImmediate()) {
-        const u32 binding_index{binding.U32()};
-        const u32 max_num_cbufs{ctx.runtime_info.max_num_cbufs};
-        if (binding_index >= max_num_cbufs) {
-            // cbuf index exceeds device limit
-            ctx.Add("{}=0u;", ret);
-            return;
-        }
-    }
     const bool is_immediate{offset.IsImmediate()};
     const bool component_indexing_bug{!is_immediate && ctx.profile.has_gl_component_indexing_bug};
     if (is_immediate) {

src/shader_recompiler/backend/glsl/glsl_emit_context.cpp

@@ -431,12 +431,6 @@ void EmitContext::DefineConstantBuffers(Bindings& bindings) {
         return;
     }
     for (const auto& desc : info.constant_buffer_descriptors) {
-        if (bindings.uniform_buffer >= runtime_info.max_num_cbufs) {
-            LOG_WARNING(Shader_GLSL, "Constant buffer binding index {} exceeds device limit of {}",
-                        bindings.uniform_buffer, runtime_info.max_num_cbufs);
-            bindings.uniform_buffer += desc.count;
-            continue;
-        }
         const auto cbuf_type{profile.has_gl_cbuf_ftou_bug ? "uvec4" : "vec4"};
         const u32 cbuf_used_size{Common::DivCeil(info.constant_buffer_used_sizes[desc.index], 16U)};
         const u32 cbuf_binding_size{info.uses_global_memory ? 0x1000U : cbuf_used_size};

src/shader_recompiler/backend/spirv/emit_spirv.h

@@ -38,10 +38,4 @@ constexpr u32 RENDERAREA_LAYOUT_OFFSET = offsetof(RenderAreaLayout, render_area)
     return EmitSPIRV(profile, {}, program, binding);
 }
 
-[[nodiscard]] inline std::vector<u32> EmitSPIRV(const Profile& profile,
-                                                const RuntimeInfo& runtime_info,
-                                                IR::Program& program) {
-    Bindings binding;
-    return EmitSPIRV(profile, runtime_info, program, binding);
-}
 } // namespace Shader::Backend::SPIRV

src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp

@@ -122,24 +122,25 @@ Id GetCbuf(EmitContext& ctx, Id result_type, Id UniformDefinitions::*member_ptr,
     if (!binding.IsImmediate()) {
         return ctx.OpFunctionCall(result_type, indirect_func, ctx.Def(binding), buffer_offset);
     }
-    const bool is_float{UniformDefinitions::IsFloat(member_ptr)};
+
-    const Id zero_val{is_float ? ctx.Const(0.0f) : ctx.Const(0u)};
+    const Id cbuf{ctx.cbufs[binding.U32()].*member_ptr};
-    const u32 binding_index{binding.U32()};
-    const u32 max_num_cbufs{ctx.runtime_info.max_num_cbufs};
-    if (binding_index >= max_num_cbufs) {
-        // cbuf index exceeds device limit
-        return zero_val;
-    }
-    const Id cbuf{ctx.cbufs[binding_index].*member_ptr};
     const Id access_chain{ctx.OpAccessChain(uniform_type, cbuf, ctx.u32_zero_value, buffer_offset)};
-    const Id val{ctx.OpLoad(result_type, access_chain)};
+    const Id val = ctx.OpLoad(result_type, access_chain);
+
     if (offset.IsImmediate() || !ctx.profile.has_broken_robust) {
         return val;
     }
-    const auto num_elements{UniformDefinitions::NumElements(member_ptr)};
+
-    const std::array zero_vec{zero_val, zero_val, zero_val, zero_val};
+    const auto is_float = UniformDefinitions::IsFloat(member_ptr);
-    const Id cond{ctx.OpULessThanEqual(ctx.TypeBool(), buffer_offset, ctx.Const(0xFFFFu))};
+    const auto num_elements = UniformDefinitions::NumElements(member_ptr);
-    const Id zero{ctx.OpCompositeConstruct(result_type, std::span(zero_vec.data(), num_elements))};
+    const std::array zero_vec{
+        is_float ? ctx.Const(0.0f) : ctx.Const(0u),
+        is_float ? ctx.Const(0.0f) : ctx.Const(0u),
+        is_float ? ctx.Const(0.0f) : ctx.Const(0u),
+        is_float ? ctx.Const(0.0f) : ctx.Const(0u),
+    };
+    const Id cond = ctx.OpULessThanEqual(ctx.TypeBool(), buffer_offset, ctx.Const(0xFFFFu));
+    const Id zero = ctx.OpCompositeConstruct(result_type, std::span(zero_vec.data(), num_elements));
     return ctx.OpSelect(result_type, cond, val, zero);
 }
 

src/shader_recompiler/backend/spirv/spirv_emit_context.cpp

@@ -278,12 +278,6 @@ void DefineConstBuffers(EmitContext& ctx, const Info& info, Id UniformDefinition
     ctx.uniform_types.*member_type = uniform_type;
 
     for (const ConstantBufferDescriptor& desc : info.constant_buffer_descriptors) {
-        if (desc.index + desc.count > ctx.runtime_info.max_num_cbufs) {
-            LOG_WARNING(Shader_SPIRV, "Constant buffer binding index {} exceeds device limit of {}",
-                        desc.index, ctx.runtime_info.max_num_cbufs);
-            binding += desc.count;
-            continue;
-        }
         const Id id{ctx.AddGlobalVariable(struct_pointer_type, spv::StorageClass::Uniform)};
         ctx.Decorate(id, spv::Decoration::Binding, binding);
         ctx.Decorate(id, spv::Decoration::DescriptorSet, 0U);

src/shader_recompiler/runtime_info.h

@@ -62,8 +62,8 @@ struct TransformFeedbackVarying {
 
 struct RuntimeInfo {
     std::array<AttributeType, 32> generic_input_types{};
-    VaryingState previous_stage_stores{};
+    VaryingState previous_stage_stores;
-    std::map<IR::Attribute, IR::Attribute> previous_stage_legacy_stores_mapping{};
+    std::map<IR::Attribute, IR::Attribute> previous_stage_legacy_stores_mapping;
 
     bool convert_depth_mode{};
     bool force_early_z{};
@@ -74,8 +74,8 @@ struct RuntimeInfo {
 
     InputTopology input_topology{};
 
-    std::optional<float> fixed_state_point_size{};
+    std::optional<float> fixed_state_point_size;
-    std::optional<CompareFunction> alpha_test_func{};
+    std::optional<CompareFunction> alpha_test_func;
     float alpha_test_reference{};
 
     /// Static Y negate value
@@ -86,9 +86,6 @@ struct RuntimeInfo {
     /// Transform feedback state for each varying
     std::array<TransformFeedbackVarying, 256> xfb_varyings{};
     u32 xfb_count{0};
-
-    /// Maximum number of UBO/CBUF bindings allowed by the host device
-    u32 max_num_cbufs{32};
 };
 
 } // namespace Shader

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -72,8 +72,7 @@ Shader::OutputTopology MaxwellToOutputTopology(Maxwell::PrimitiveTopology topolo
 Shader::RuntimeInfo MakeRuntimeInfo(const GraphicsPipelineKey& key,
                                     const Shader::IR::Program& program,
                                     const Shader::IR::Program* previous_program,
-                                    bool glasm_use_storage_buffers, bool use_assembly_shaders,
+                                    bool glasm_use_storage_buffers, bool use_assembly_shaders) {
-                                    u32 max_num_cbufs) {
     Shader::RuntimeInfo info;
     if (previous_program) {
         info.previous_stage_stores = previous_program->info.stores;
@@ -153,7 +152,6 @@ Shader::RuntimeInfo MakeRuntimeInfo(const GraphicsPipelineKey& key,
         break;
     }
     info.glasm_use_storage_buffers = glasm_use_storage_buffers;
-    info.max_num_cbufs = max_num_cbufs;
     return info;
 }
 
@@ -524,9 +522,8 @@ std::unique_ptr<GraphicsPipeline> ShaderCache::CreateGraphicsPipeline(
         const size_t stage_index{index - 1};
         infos[stage_index] = &program.info;
 
-        const u32 max_num_cbufs{device.GetMaxUniformBuffers(program.stage)};
+        const auto runtime_info{
-        const auto runtime_info{MakeRuntimeInfo(
+            MakeRuntimeInfo(key, program, previous_program, glasm_use_storage_buffers, use_glasm)};
-            key, program, previous_program, glasm_use_storage_buffers, use_glasm, max_num_cbufs)};
         switch (device.GetShaderBackend()) {
         case Settings::ShaderBackend::Glsl:
             ConvertLegacyToGeneric(program, runtime_info);
@@ -583,21 +580,20 @@ std::unique_ptr<ComputePipeline> ShaderCache::CreateComputePipeline(
 
     auto program{TranslateProgram(pools.inst, pools.block, env, cfg, host_info)};
     const u32 num_storage_buffers{Shader::NumDescriptors(program.info.storage_buffers_descriptors)};
-    const Shader::RuntimeInfo info{
+    Shader::RuntimeInfo info;
-        .glasm_use_storage_buffers = num_storage_buffers <= device.GetMaxGLASMStorageBufferBlocks(),
+    info.glasm_use_storage_buffers = num_storage_buffers <= device.GetMaxGLASMStorageBufferBlocks();
-        .max_num_cbufs = device.GetMaxUniformBuffers(program.stage),
+
-    };
     std::string code{};
     std::vector<u32> code_spirv;
     switch (device.GetShaderBackend()) {
     case Settings::ShaderBackend::Glsl:
-        code = EmitGLSL(profile, info, program);
+        code = EmitGLSL(profile, program);
         break;
     case Settings::ShaderBackend::Glasm:
         code = EmitGLASM(profile, info, program);
         break;
     case Settings::ShaderBackend::SpirV:
-        code_spirv = EmitSPIRV(profile, info, program);
+        code_spirv = EmitSPIRV(profile, program);
         break;
     }
 

src/video_core/renderer_vulkan/vk_pipeline_cache.cpp

@@ -145,8 +145,7 @@ Shader::AttributeType AttributeType(const FixedPipelineState& state, size_t inde
 Shader::RuntimeInfo MakeRuntimeInfo(std::span<const Shader::IR::Program> programs,
                                     const GraphicsPipelineCacheKey& key,
                                     const Shader::IR::Program& program,
-                                    const Shader::IR::Program* previous_program,
+                                    const Shader::IR::Program* previous_program) {
-                                    u32 max_num_cbufs) {
     Shader::RuntimeInfo info;
     if (previous_program) {
         info.previous_stage_stores = previous_program->info.stores;
@@ -262,7 +261,6 @@ Shader::RuntimeInfo MakeRuntimeInfo(std::span<const Shader::IR::Program> program
     }
     info.force_early_z = key.state.early_z != 0;
     info.y_negate = key.state.y_negate != 0;
-    info.max_num_cbufs = max_num_cbufs;
     return info;
 }
 
@@ -657,7 +655,6 @@ std::unique_ptr<GraphicsPipeline> PipelineCache::CreateGraphicsPipeline(
 
     const Shader::IR::Program* previous_stage{};
     Shader::Backend::Bindings binding;
-    const u32 max_num_cbufs{static_cast<u32>(device.GetMaxPerStageUniformBuffers())};
     for (size_t index = uses_vertex_a && uses_vertex_b ? 1 : 0; index < Maxwell::MaxShaderProgram;
          ++index) {
         const bool is_emulated_stage = layer_source_program != nullptr &&
@@ -671,8 +668,7 @@ std::unique_ptr<GraphicsPipeline> PipelineCache::CreateGraphicsPipeline(
         const size_t stage_index{index - 1};
         infos[stage_index] = &program.info;
 
-        const auto runtime_info{
+        const auto runtime_info{MakeRuntimeInfo(programs, key, program, previous_stage)};
-            MakeRuntimeInfo(programs, key, program, previous_stage, max_num_cbufs)};
         ConvertLegacyToGeneric(program, runtime_info);
         const std::vector<u32> code{EmitSPIRV(profile, runtime_info, program, binding)};
         device.SaveShader(code);
@@ -768,10 +764,7 @@ std::unique_ptr<ComputePipeline> PipelineCache::CreateComputePipeline(
     }
 
     auto program{TranslateProgram(pools.inst, pools.block, env, cfg, host_info)};
-    const Shader::RuntimeInfo info{
+    const std::vector<u32> code{EmitSPIRV(profile, program)};
-        .max_num_cbufs = static_cast<u32>(device.GetMaxPerStageUniformBuffers()),
-    };
-    const std::vector<u32> code{EmitSPIRV(profile, info, program)};
     device.SaveShader(code);
     vk::ShaderModule spv_module{BuildShader(device, code)};
     if (device.HasDebuggingToolAttached()) {

src/video_core/vulkan_common/vulkan_device.h

@@ -299,11 +299,6 @@ public:
         return properties.properties.limits.maxComputeSharedMemorySize;
     }
 
-    /// Returns the maximum number of uniform buffers allowed per stage.
-    VkDeviceSize GetMaxPerStageUniformBuffers() const {
-        return properties.properties.limits.maxPerStageDescriptorUniformBuffers;
-    }
-
     /// Returns float control properties of the device.
     const VkPhysicalDeviceFloatControlsPropertiesKHR& FloatControlProperties() const {
         return properties.float_controls;