renderer_vulkan: skip SetObjectNameEXT on unsupported driver

OpenGL: Add GL_PRIMITIVES_GENERATED and GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN queries

src/shader_recompiler/backend/spirv/emit_spirv_image.cpp

@@ -214,16 +214,16 @@ Id TextureImage(EmitContext& ctx, IR::TextureInstInfo info, const IR::Value& ind
     }
 }
 
-Id Image(EmitContext& ctx, const IR::Value& index, IR::TextureInstInfo info) {
+std::pair<Id, bool> Image(EmitContext& ctx, const IR::Value& index, IR::TextureInstInfo info) {
     if (!index.IsImmediate() || index.U32() != 0) {
         throw NotImplementedException("Indirect image indexing");
     }
     if (info.type == TextureType::Buffer) {
         const ImageBufferDefinition def{ctx.image_buffers.at(info.descriptor_index)};
-        return ctx.OpLoad(def.image_type, def.id);
+        return {ctx.OpLoad(def.image_type, def.id), def.is_integer};
     } else {
         const ImageDefinition def{ctx.images.at(info.descriptor_index)};
-        return ctx.OpLoad(def.image_type, def.id);
+        return {ctx.OpLoad(def.image_type, def.id), def.is_integer};
     }
 }
 
@@ -566,13 +566,23 @@ Id EmitImageRead(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id co
         LOG_WARNING(Shader_SPIRV, "Typeless image read not supported by host");
         return ctx.ConstantNull(ctx.U32[4]);
     }
-    return Emit(&EmitContext::OpImageSparseRead, &EmitContext::OpImageRead, ctx, inst, ctx.U32[4],
-                Image(ctx, index, info), coords, std::nullopt, std::span<const Id>{});
+    const auto [image, is_integer] = Image(ctx, index, info);
+    const Id result_type{is_integer ? ctx.U32[4] : ctx.F32[4]};
+    Id color{Emit(&EmitContext::OpImageSparseRead, &EmitContext::OpImageRead, ctx, inst,
+                  result_type, image, coords, std::nullopt, std::span<const Id>{})};
+    if (!is_integer) {
+        color = ctx.OpBitcast(ctx.U32[4], color);
+    }
+    return color;
 }
 
 void EmitImageWrite(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id coords, Id color) {
     const auto info{inst->Flags<IR::TextureInstInfo>()};
-    ctx.OpImageWrite(Image(ctx, index, info), coords, color);
+    const auto [image, is_integer] = Image(ctx, index, info);
+    if (!is_integer) {
+        color = ctx.OpBitcast(ctx.F32[4], color);
+    }
+    ctx.OpImageWrite(image, coords, color);
 }
 
 Id EmitIsTextureScaled(EmitContext& ctx, const IR::Value& index) {

src/shader_recompiler/backend/spirv/spirv_emit_context.cpp

@@ -74,28 +74,19 @@ spv::ImageFormat GetImageFormat(ImageFormat format) {
     throw InvalidArgument("Invalid image format {}", format);
 }
 
-Id GetImageSampledType(EmitContext& ctx, const ImageDescriptor& desc) {
-    if (desc.is_float) {
-        return ctx.F32[1];
-    } else {
-        return ctx.U32[1];
-    }
-}
-
-Id ImageType(EmitContext& ctx, const ImageDescriptor& desc) {
+Id ImageType(EmitContext& ctx, const ImageDescriptor& desc, Id sampled_type) {
     const spv::ImageFormat format{GetImageFormat(desc.format)};
-    const Id type{GetImageSampledType(ctx, desc)};
     switch (desc.type) {
     case TextureType::Color1D:
-        return ctx.TypeImage(type, spv::Dim::Dim1D, false, false, false, 2, format);
+        return ctx.TypeImage(sampled_type, spv::Dim::Dim1D, false, false, false, 2, format);
     case TextureType::ColorArray1D:
-        return ctx.TypeImage(type, spv::Dim::Dim1D, false, true, false, 2, format);
+        return ctx.TypeImage(sampled_type, spv::Dim::Dim1D, false, true, false, 2, format);
     case TextureType::Color2D:
-        return ctx.TypeImage(type, spv::Dim::Dim2D, false, false, false, 2, format);
+        return ctx.TypeImage(sampled_type, spv::Dim::Dim2D, false, false, false, 2, format);
     case TextureType::ColorArray2D:
-        return ctx.TypeImage(type, spv::Dim::Dim2D, false, true, false, 2, format);
+        return ctx.TypeImage(sampled_type, spv::Dim::Dim2D, false, true, false, 2, format);
     case TextureType::Color3D:
-        return ctx.TypeImage(type, spv::Dim::Dim3D, false, false, false, 2, format);
+        return ctx.TypeImage(sampled_type, spv::Dim::Dim3D, false, false, false, 2, format);
     case TextureType::Buffer:
         throw NotImplementedException("Image buffer");
     default:
@@ -1281,7 +1272,9 @@ void EmitContext::DefineImageBuffers(const Info& info, u32& binding) {
             throw NotImplementedException("Array of image buffers");
         }
         const spv::ImageFormat format{GetImageFormat(desc.format)};
-        const Id image_type{TypeImage(U32[1], spv::Dim::Buffer, false, false, false, 2, format)};
+        const Id sampled_type{desc.is_integer ? U32[1] : F32[1]};
+        const Id image_type{
+            TypeImage(sampled_type, spv::Dim::Buffer, false, false, false, 2, format)};
         const Id pointer_type{TypePointer(spv::StorageClass::UniformConstant, image_type)};
         const Id id{AddGlobalVariable(pointer_type, spv::StorageClass::UniformConstant)};
         Decorate(id, spv::Decoration::Binding, binding);
@@ -1291,6 +1284,7 @@ void EmitContext::DefineImageBuffers(const Info& info, u32& binding) {
             .id = id,
             .image_type = image_type,
             .count = desc.count,
+            .is_integer = desc.is_integer,
         });
         if (profile.supported_spirv >= 0x00010400) {
             interfaces.push_back(id);
@@ -1335,7 +1329,8 @@ void EmitContext::DefineImages(const Info& info, u32& binding, u32& scaling_inde
         if (desc.count != 1) {
             throw NotImplementedException("Array of images");
         }
-        const Id image_type{ImageType(*this, desc)};
+        const Id sampled_type{desc.is_integer ? U32[1] : F32[1]};
+        const Id image_type{ImageType(*this, desc, sampled_type)};
         const Id pointer_type{TypePointer(spv::StorageClass::UniformConstant, image_type)};
         const Id id{AddGlobalVariable(pointer_type, spv::StorageClass::UniformConstant)};
         Decorate(id, spv::Decoration::Binding, binding);
@@ -1345,6 +1340,7 @@ void EmitContext::DefineImages(const Info& info, u32& binding, u32& scaling_inde
             .id = id,
             .image_type = image_type,
             .count = desc.count,
+            .is_integer = desc.is_integer,
         });
         if (profile.supported_spirv >= 0x00010400) {
             interfaces.push_back(id);

src/shader_recompiler/backend/spirv/spirv_emit_context.h

@@ -47,12 +47,14 @@ struct ImageBufferDefinition {
     Id id;
     Id image_type;
     u32 count;
+    bool is_integer;
 };
 
 struct ImageDefinition {
     Id id;
     Id image_type;
     u32 count;
+    bool is_integer;
 };
 
 struct UniformDefinitions {

src/shader_recompiler/environment.h

@@ -24,6 +24,8 @@ public:
 
     [[nodiscard]] virtual TexturePixelFormat ReadTexturePixelFormat(u32 raw_handle) = 0;
 
+    [[nodiscard]] virtual bool IsTexturePixelFormatInteger(u32 raw_handle) = 0;
+
     [[nodiscard]] virtual u32 ReadViewportTransformState() = 0;
 
     [[nodiscard]] virtual u32 TextureBoundBuffer() const = 0;

src/shader_recompiler/frontend/ir/opcodes.inc

@@ -512,7 +512,7 @@ OPCODE(ImageQueryDimensions,                                U32x4,          Opaq
 OPCODE(ImageQueryLod,                                       F32x4,          Opaque,         Opaque,                                                         )
 OPCODE(ImageGradient,                                       F32x4,          Opaque,         Opaque,         Opaque,         Opaque,         Opaque,         )
 OPCODE(ImageRead,                                           U32x4,          Opaque,         Opaque,                                                         )
-OPCODE(ImageWrite,                                          Void,           Opaque,         Opaque,         Opaque,                                         )
+OPCODE(ImageWrite,                                          Void,           Opaque,         Opaque,         U32x4,                                          )
 
 OPCODE(IsTextureScaled,                                     U1,             U32,                                                                            )
 OPCODE(IsImageScaled,                                       U1,             U32,                                                                            )

src/shader_recompiler/host_translate_info.h

@@ -19,10 +19,8 @@ struct HostTranslateInfo {
     u32 min_ssbo_alignment{};            ///< Minimum alignment supported by the device for SSBOs
     bool support_geometry_shader_passthrough{}; ///< True when the device supports geometry
                                                 ///< passthrough shaders
-    bool support_conditional_barrier{};  ///< True when the device supports barriers in conditional
-                                         ///< control flow
-    bool support_ufloat_write_as_uint{}; ///< True when the device supports writing float images
-                                         ///< as bitcasts to uint
+    bool support_conditional_barrier{}; ///< True when the device supports barriers in conditional
+                                        ///< control flow
 };
 
 } // namespace Shader

src/shader_recompiler/ir_opt/constant_propagation_pass.cpp

@@ -815,6 +815,15 @@ bool FindGradient3DDerivatives(std::array<IR::Value, 3>& results, IR::Value coor
     return true;
 }
 
+void ConvertDerivatives(std::array<IR::Value, 3>& results, IR::IREmitter& ir) {
+    for (size_t i = 0; i < 3; i++) {
+        if (results[i].Type() == IR::Type::U32) {
+            results[i] = results[i].IsImmediate() ? ir.Imm32(Common::BitCast<f32>(results[i].U32()))
+                                                  : ir.BitCast<IR::F32>(IR::U32(results[i]));
+        }
+    }
+}
+
 void FoldImageSampleImplicitLod(IR::Block& block, IR::Inst& inst) {
     IR::TextureInstInfo info = inst.Flags<IR::TextureInstInfo>();
     auto orig_opcode = inst.GetOpcode();
@@ -831,12 +840,14 @@ void FoldImageSampleImplicitLod(IR::Block& block, IR::Inst& inst) {
     if (!offset.IsImmediate()) {
         return;
     }
+    IR::IREmitter ir{block, IR::Block::InstructionList::s_iterator_to(inst)};
     IR::Inst* const inst2 = coords.InstRecursive();
     std::array<std::array<IR::Value, 3>, 3> results_matrix;
     for (size_t i = 0; i < 3; i++) {
         if (!FindGradient3DDerivatives(results_matrix[i], inst2->Arg(i).Resolve())) {
             return;
         }
+        ConvertDerivatives(results_matrix[i], ir);
     }
     IR::F32 lod_clamp{};
     if (info.has_lod_clamp != 0) {
@@ -846,7 +857,6 @@ void FoldImageSampleImplicitLod(IR::Block& block, IR::Inst& inst) {
             lod_clamp = IR::F32{bias_lc};
         }
     }
-    IR::IREmitter ir{block, IR::Block::InstructionList::s_iterator_to(inst)};
     IR::Value new_coords =
         ir.CompositeConstruct(results_matrix[0][0], results_matrix[1][0], results_matrix[2][0]);
     IR::Value derivatives_1 = ir.CompositeConstruct(results_matrix[0][1], results_matrix[0][2],

src/shader_recompiler/ir_opt/texture_pass.cpp

@@ -372,8 +372,8 @@ TexturePixelFormat ReadTexturePixelFormat(Environment& env, const ConstBufferAdd
     return env.ReadTexturePixelFormat(GetTextureHandle(env, cbuf));
 }
 
-bool TexturePixelFormatIsFloat(Environment& env, const ConstBufferAddr& cbuf) {
-    return ReadTexturePixelFormat(env, cbuf) == TexturePixelFormat::B10G11R11_FLOAT;
+bool IsTexturePixelFormatInteger(Environment& env, const ConstBufferAddr& cbuf) {
+    return env.IsTexturePixelFormatInteger(GetTextureHandle(env, cbuf));
 }
 
 class Descriptors {
@@ -407,6 +407,7 @@ public:
         })};
         image_buffer_descriptors[index].is_written |= desc.is_written;
         image_buffer_descriptors[index].is_read |= desc.is_read;
+        image_buffer_descriptors[index].is_integer |= desc.is_integer;
         return index;
     }
 
@@ -432,11 +433,11 @@ public:
             return desc.type == existing.type && desc.format == existing.format &&
                    desc.cbuf_index == existing.cbuf_index &&
                    desc.cbuf_offset == existing.cbuf_offset && desc.count == existing.count &&
-                   desc.size_shift == existing.size_shift && desc.is_float == existing.is_float;
+                   desc.size_shift == existing.size_shift;
         })};
-        // TODO: handle is_float?
         image_descriptors[index].is_written |= desc.is_written;
         image_descriptors[index].is_read |= desc.is_read;
+        image_descriptors[index].is_integer |= desc.is_integer;
         return index;
     }
 
@@ -474,6 +475,20 @@ void PatchImageSampleImplicitLod(IR::Block& block, IR::Inst& inst) {
                         ir.FPRecip(ir.ConvertUToF(32, 32, ir.CompositeExtract(texture_size, 1))))));
 }
 
+bool IsPixelFormatSNorm(TexturePixelFormat pixel_format) {
+    switch (pixel_format) {
+    case TexturePixelFormat::A8B8G8R8_SNORM:
+    case TexturePixelFormat::R8G8_SNORM:
+    case TexturePixelFormat::R8_SNORM:
+    case TexturePixelFormat::R16G16B16A16_SNORM:
+    case TexturePixelFormat::R16G16_SNORM:
+    case TexturePixelFormat::R16_SNORM:
+        return true;
+    default:
+        return false;
+    }
+}
+
 void PatchTexelFetch(IR::Block& block, IR::Inst& inst, TexturePixelFormat pixel_format) {
     const auto it{IR::Block::InstructionList::s_iterator_to(inst)};
     IR::IREmitter ir{block, IR::Block::InstructionList::s_iterator_to(inst)};
@@ -505,19 +520,6 @@ void PatchTexelFetch(IR::Block& block, IR::Inst& inst, TexturePixelFormat pixel_
                               ir.FPMul(ir.ConvertSToF(32, 32, ir.BitCast<IR::U32>(w)), max_value));
     inst.ReplaceUsesWith(converted);
 }
-
-void PatchSmallFloatImageWrite(IR::Block& block, IR::Inst& inst) {
-    IR::IREmitter ir{block, IR::Block::InstructionList::s_iterator_to(inst)};
-
-    const IR::Value old_value{inst.Arg(2)};
-    const IR::F32 x(ir.BitCast<IR::F32>(IR::U32(ir.CompositeExtract(old_value, 0))));
-    const IR::F32 y(ir.BitCast<IR::F32>(IR::U32(ir.CompositeExtract(old_value, 1))));
-    const IR::F32 z(ir.BitCast<IR::F32>(IR::U32(ir.CompositeExtract(old_value, 2))));
-    const IR::F32 w(ir.BitCast<IR::F32>(IR::U32(ir.CompositeExtract(old_value, 3))));
-    const IR::Value converted = ir.CompositeConstruct(x, y, z, w);
-    inst.SetArg(2, converted);
-}
-
 } // Anonymous namespace
 
 void TexturePass(Environment& env, IR::Program& program, const HostTranslateInfo& host_info) {
@@ -549,9 +551,6 @@ void TexturePass(Environment& env, IR::Program& program, const HostTranslateInfo
         inst->ReplaceOpcode(IndexedInstruction(*inst));
 
         const auto& cbuf{texture_inst.cbuf};
-        const bool is_float_write{!host_info.support_ufloat_write_as_uint &&
-                                  inst->GetOpcode() == IR::Opcode::ImageWrite &&
-                                  TexturePixelFormatIsFloat(env, cbuf)};
         auto flags{inst->Flags<IR::TextureInstInfo>()};
         bool is_multisample{false};
         switch (inst->GetOpcode()) {
@@ -608,11 +607,13 @@ void TexturePass(Environment& env, IR::Program& program, const HostTranslateInfo
             }
             const bool is_written{inst->GetOpcode() != IR::Opcode::ImageRead};
             const bool is_read{inst->GetOpcode() != IR::Opcode::ImageWrite};
+            const bool is_integer{IsTexturePixelFormatInteger(env, cbuf)};
             if (flags.type == TextureType::Buffer) {
                 index = descriptors.Add(ImageBufferDescriptor{
                     .format = flags.image_format,
                     .is_written = is_written,
                     .is_read = is_read,
+                    .is_integer = is_integer,
                     .cbuf_index = cbuf.index,
                     .cbuf_offset = cbuf.offset,
                     .count = cbuf.count,
@@ -624,7 +625,7 @@ void TexturePass(Environment& env, IR::Program& program, const HostTranslateInfo
                     .format = flags.image_format,
                     .is_written = is_written,
                     .is_read = is_read,
-                    .is_float = is_float_write,
+                    .is_integer = is_integer,
                     .cbuf_index = cbuf.index,
                     .cbuf_offset = cbuf.offset,
                     .count = cbuf.count,
@@ -680,14 +681,10 @@ void TexturePass(Environment& env, IR::Program& program, const HostTranslateInfo
         if (!host_info.support_snorm_render_buffer && inst->GetOpcode() == IR::Opcode::ImageFetch &&
             flags.type == TextureType::Buffer) {
             const auto pixel_format = ReadTexturePixelFormat(env, cbuf);
-            if (pixel_format != TexturePixelFormat::OTHER) {
+            if (IsPixelFormatSNorm(pixel_format)) {
                 PatchTexelFetch(*texture_inst.block, *texture_inst.inst, pixel_format);
             }
         }
-
-        if (is_float_write) {
-            PatchSmallFloatImageWrite(*texture_inst.block, *inst);
-        }
     }
 }
 

src/shader_recompiler/shader_info.h

@@ -35,15 +35,109 @@ enum class TextureType : u32 {
 };
 constexpr u32 NUM_TEXTURE_TYPES = 9;
 
-enum class TexturePixelFormat : u32 {
+enum class TexturePixelFormat {
+    A8B8G8R8_UNORM,
     A8B8G8R8_SNORM,
+    A8B8G8R8_SINT,
+    A8B8G8R8_UINT,
+    R5G6B5_UNORM,
+    B5G6R5_UNORM,
+    A1R5G5B5_UNORM,
+    A2B10G10R10_UNORM,
+    A2B10G10R10_UINT,
+    A2R10G10B10_UNORM,
+    A1B5G5R5_UNORM,
+    A5B5G5R1_UNORM,
+    R8_UNORM,
     R8_SNORM,
-    R8G8_SNORM,
+    R8_SINT,
+    R8_UINT,
+    R16G16B16A16_FLOAT,
+    R16G16B16A16_UNORM,
     R16G16B16A16_SNORM,
-    R16G16_SNORM,
-    R16_SNORM,
+    R16G16B16A16_SINT,
+    R16G16B16A16_UINT,
     B10G11R11_FLOAT,
-    OTHER
+    R32G32B32A32_UINT,
+    BC1_RGBA_UNORM,
+    BC2_UNORM,
+    BC3_UNORM,
+    BC4_UNORM,
+    BC4_SNORM,
+    BC5_UNORM,
+    BC5_SNORM,
+    BC7_UNORM,
+    BC6H_UFLOAT,
+    BC6H_SFLOAT,
+    ASTC_2D_4X4_UNORM,
+    B8G8R8A8_UNORM,
+    R32G32B32A32_FLOAT,
+    R32G32B32A32_SINT,
+    R32G32_FLOAT,
+    R32G32_SINT,
+    R32_FLOAT,
+    R16_FLOAT,
+    R16_UNORM,
+    R16_SNORM,
+    R16_UINT,
+    R16_SINT,
+    R16G16_UNORM,
+    R16G16_FLOAT,
+    R16G16_UINT,
+    R16G16_SINT,
+    R16G16_SNORM,
+    R32G32B32_FLOAT,
+    A8B8G8R8_SRGB,
+    R8G8_UNORM,
+    R8G8_SNORM,
+    R8G8_SINT,
+    R8G8_UINT,
+    R32G32_UINT,
+    R16G16B16X16_FLOAT,
+    R32_UINT,
+    R32_SINT,
+    ASTC_2D_8X8_UNORM,
+    ASTC_2D_8X5_UNORM,
+    ASTC_2D_5X4_UNORM,
+    B8G8R8A8_SRGB,
+    BC1_RGBA_SRGB,
+    BC2_SRGB,
+    BC3_SRGB,
+    BC7_SRGB,
+    A4B4G4R4_UNORM,
+    G4R4_UNORM,
+    ASTC_2D_4X4_SRGB,
+    ASTC_2D_8X8_SRGB,
+    ASTC_2D_8X5_SRGB,
+    ASTC_2D_5X4_SRGB,
+    ASTC_2D_5X5_UNORM,
+    ASTC_2D_5X5_SRGB,
+    ASTC_2D_10X8_UNORM,
+    ASTC_2D_10X8_SRGB,
+    ASTC_2D_6X6_UNORM,
+    ASTC_2D_6X6_SRGB,
+    ASTC_2D_10X6_UNORM,
+    ASTC_2D_10X6_SRGB,
+    ASTC_2D_10X5_UNORM,
+    ASTC_2D_10X5_SRGB,
+    ASTC_2D_10X10_UNORM,
+    ASTC_2D_10X10_SRGB,
+    ASTC_2D_12X10_UNORM,
+    ASTC_2D_12X10_SRGB,
+    ASTC_2D_12X12_UNORM,
+    ASTC_2D_12X12_SRGB,
+    ASTC_2D_8X6_UNORM,
+    ASTC_2D_8X6_SRGB,
+    ASTC_2D_6X5_UNORM,
+    ASTC_2D_6X5_SRGB,
+    E5B9G9R9_FLOAT,
+    D32_FLOAT,
+    D16_UNORM,
+    X8_D24_UNORM,
+    S8_UINT,
+    D24_UNORM_S8_UINT,
+    S8_UINT_D24_UNORM,
+    D32_FLOAT_S8_UINT,
 };
 
 enum class ImageFormat : u32 {
@@ -98,6 +192,7 @@ struct ImageBufferDescriptor {
     ImageFormat format;
     bool is_written;
     bool is_read;
+    bool is_integer;
     u32 cbuf_index;
     u32 cbuf_offset;
     u32 count;
@@ -130,7 +225,7 @@ struct ImageDescriptor {
     ImageFormat format;
     bool is_written;
     bool is_read;
-    bool is_float;
+    bool is_integer;
     u32 cbuf_index;
     u32 cbuf_offset;
     u32 count;

src/video_core/engines/maxwell_3d.cpp

@@ -586,14 +586,22 @@ void Maxwell3D::ProcessQueryCondition() {
 }
 
 void Maxwell3D::ProcessCounterReset() {
-    switch (regs.clear_report_value) {
-    case Regs::ClearReport::ZPassPixelCount:
-        rasterizer->ResetCounter(VideoCommon::QueryType::ZPassPixelCount64);
-        break;
-    default:
-        LOG_DEBUG(Render_OpenGL, "Unimplemented counter reset={}", regs.clear_report_value);
-        break;
-    }
+    const auto query_type = [clear_report = regs.clear_report_value]() {
+        switch (clear_report) {
+        case Tegra::Engines::Maxwell3D::Regs::ClearReport::ZPassPixelCount:
+            return VideoCommon::QueryType::ZPassPixelCount64;
+        case Tegra::Engines::Maxwell3D::Regs::ClearReport::StreamingPrimitivesSucceeded:
+            return VideoCommon::QueryType::StreamingPrimitivesSucceeded;
+        case Tegra::Engines::Maxwell3D::Regs::ClearReport::PrimitivesGenerated:
+            return VideoCommon::QueryType::PrimitivesGenerated;
+        case Tegra::Engines::Maxwell3D::Regs::ClearReport::VtgPrimitivesOut:
+            return VideoCommon::QueryType::VtgPrimitivesOut;
+        default:
+            LOG_DEBUG(HW_GPU, "Unimplemented counter reset={}", clear_report);
+            return VideoCommon::QueryType::Payload;
+        }
+    }();
+    rasterizer->ResetCounter(query_type);
 }
 
 void Maxwell3D::ProcessSyncPoint() {

src/video_core/query_cache.h

@@ -28,8 +28,11 @@
 namespace VideoCore {
 enum class QueryType {
     SamplesPassed,
+    PrimitivesGenerated,
+    TfbPrimitivesWritten,
+    Count,
 };
-constexpr std::size_t NumQueryTypes = 1;
+constexpr std::size_t NumQueryTypes = static_cast<size_t>(QueryType::Count);
 } // namespace VideoCore
 
 namespace VideoCommon {
@@ -44,15 +47,6 @@ public:
     explicit CounterStreamBase(QueryCache& cache_, VideoCore::QueryType type_)
         : cache{cache_}, type{type_} {}
 
-    /// Updates the state of the stream, enabling or disabling as needed.
-    void Update(bool enabled) {
-        if (enabled) {
-            Enable();
-        } else {
-            Disable();
-        }
-    }
-
     /// Resets the stream to zero. It doesn't disable the query after resetting.
     void Reset() {
         if (current) {
@@ -80,7 +74,6 @@ public:
         return current != nullptr;
     }
 
-private:
     /// Enables the stream.
     void Enable() {
         if (current) {
@@ -97,6 +90,7 @@ private:
         last = std::exchange(current, nullptr);
     }
 
+private:
     QueryCache& cache;
     const VideoCore::QueryType type;
 
@@ -112,8 +106,14 @@ public:
         : rasterizer{rasterizer_},
           // Use reinterpret_cast instead of static_cast as workaround for
           // UBSan bug (https://github.com/llvm/llvm-project/issues/59060)
-          cpu_memory{cpu_memory_}, streams{{CounterStream{reinterpret_cast<QueryCache&>(*this),
-                                                          VideoCore::QueryType::SamplesPassed}}} {
+          cpu_memory{cpu_memory_}, streams{{
+                                       {CounterStream{reinterpret_cast<QueryCache&>(*this),
+                                                      VideoCore::QueryType::SamplesPassed}},
+                                       {CounterStream{reinterpret_cast<QueryCache&>(*this),
+                                                      VideoCore::QueryType::PrimitivesGenerated}},
+                                       {CounterStream{reinterpret_cast<QueryCache&>(*this),
+                                                      VideoCore::QueryType::TfbPrimitivesWritten}},
+                                   }} {
         (void)slot_async_jobs.insert(); // Null value
     }
 
@@ -157,12 +157,11 @@ public:
         AsyncFlushQuery(query, timestamp, lock);
     }
 
-    /// Updates counters from GPU state. Expected to be called once per draw, clear or dispatch.
-    void UpdateCounters() {
+    /// Enables all available GPU counters
+    void EnableCounters() {
         std::unique_lock lock{mutex};
-        if (maxwell3d) {
-            const auto& regs = maxwell3d->regs;
-            Stream(VideoCore::QueryType::SamplesPassed).Update(regs.zpass_pixel_count_enable);
+        for (auto& stream : streams) {
+            stream.Enable();
         }
     }
 
@@ -176,7 +175,7 @@ public:
     void DisableStreams() {
         std::unique_lock lock{mutex};
         for (auto& stream : streams) {
-            stream.Update(false);
+            stream.Disable();
         }
     }
 
@@ -353,7 +352,7 @@ private:
 
     std::shared_ptr<std::vector<AsyncJobId>> uncommitted_flushes{};
     std::list<std::shared_ptr<std::vector<AsyncJobId>>> committed_flushes;
-};
+}; // namespace VideoCommon
 
 template <class QueryCache, class HostCounter>
 class HostCounterBase {

src/video_core/renderer_opengl/gl_query_cache.cpp

@@ -18,16 +18,27 @@ namespace OpenGL {
 
 namespace {
 
-constexpr std::array<GLenum, VideoCore::NumQueryTypes> QueryTargets = {GL_SAMPLES_PASSED};
-
 constexpr GLenum GetTarget(VideoCore::QueryType type) {
-    return QueryTargets[static_cast<std::size_t>(type)];
+    switch (type) {
+    case VideoCore::QueryType::SamplesPassed:
+        return GL_SAMPLES_PASSED;
+    case VideoCore::QueryType::PrimitivesGenerated:
+        return GL_PRIMITIVES_GENERATED;
+    case VideoCore::QueryType::TfbPrimitivesWritten:
+        return GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN;
+    default:
+        break;
+    }
+    UNIMPLEMENTED_MSG("Query type {}", type);
+    return 0;
 }
 
 } // Anonymous namespace
 
 QueryCache::QueryCache(RasterizerOpenGL& rasterizer_, Core::Memory::Memory& cpu_memory_)
-    : QueryCacheLegacy(rasterizer_, cpu_memory_), gl_rasterizer{rasterizer_} {}
+    : QueryCacheLegacy(rasterizer_, cpu_memory_), gl_rasterizer{rasterizer_} {
+    EnableCounters();
+}
 
 QueryCache::~QueryCache() = default;
 
@@ -103,13 +114,13 @@ u64 CachedQuery::Flush([[maybe_unused]] bool async) {
     auto& stream = cache->Stream(type);
     const bool slice_counter = WaitPending() && stream.IsEnabled();
     if (slice_counter) {
-        stream.Update(false);
+        stream.Disable();
     }
 
     auto result = VideoCommon::CachedQueryBase<HostCounter>::Flush();
 
     if (slice_counter) {
-        stream.Update(true);
+        stream.Enable();
     }
 
     return result;

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -51,6 +51,22 @@ constexpr size_t NUM_SUPPORTED_VERTEX_ATTRIBUTES = 16;
 void oglEnable(GLenum cap, bool state) {
     (state ? glEnable : glDisable)(cap);
 }
+
+std::optional<VideoCore::QueryType> MaxwellToVideoCoreQuery(VideoCommon::QueryType type) {
+    switch (type) {
+    case VideoCommon::QueryType::PrimitivesGenerated:
+    case VideoCommon::QueryType::VtgPrimitivesOut:
+        return VideoCore::QueryType::PrimitivesGenerated;
+    case VideoCommon::QueryType::ZPassPixelCount64:
+        return VideoCore::QueryType::SamplesPassed;
+    case VideoCommon::QueryType::StreamingPrimitivesSucceeded:
+        // case VideoCommon::QueryType::StreamingByteCount:
+        // TODO: StreamingByteCount = StreamingPrimitivesSucceeded * num_verts * vert_stride
+        return VideoCore::QueryType::TfbPrimitivesWritten;
+    default:
+        return std::nullopt;
+    }
+}
 } // Anonymous namespace
 
 RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_,
@@ -216,7 +232,6 @@ void RasterizerOpenGL::PrepareDraw(bool is_indexed, Func&& draw_func) {
 
     SCOPE_EXIT({ gpu.TickWork(); });
     gpu_memory->FlushCaching();
-    query_cache.UpdateCounters();
 
     GraphicsPipeline* const pipeline{shader_cache.CurrentGraphicsPipeline()};
     if (!pipeline) {
@@ -334,7 +349,6 @@ void RasterizerOpenGL::DrawTexture() {
     MICROPROFILE_SCOPE(OpenGL_Drawing);
 
     SCOPE_EXIT({ gpu.TickWork(); });
-    query_cache.UpdateCounters();
 
     texture_cache.SynchronizeGraphicsDescriptors();
     texture_cache.UpdateRenderTargets(false);
@@ -401,21 +415,28 @@ void RasterizerOpenGL::DispatchCompute() {
 }
 
 void RasterizerOpenGL::ResetCounter(VideoCommon::QueryType type) {
-    if (type == VideoCommon::QueryType::ZPassPixelCount64) {
-        query_cache.ResetCounter(VideoCore::QueryType::SamplesPassed);
+    const auto query_cache_type = MaxwellToVideoCoreQuery(type);
+    if (!query_cache_type.has_value()) {
+        UNIMPLEMENTED_IF_MSG(type != VideoCommon::QueryType::Payload, "Reset query type: {}", type);
+        return;
     }
+    query_cache.ResetCounter(*query_cache_type);
 }
 
 void RasterizerOpenGL::Query(GPUVAddr gpu_addr, VideoCommon::QueryType type,
                              VideoCommon::QueryPropertiesFlags flags, u32 payload, u32 subreport) {
-    if (type == VideoCommon::QueryType::ZPassPixelCount64) {
-        if (True(flags & VideoCommon::QueryPropertiesFlags::HasTimeout)) {
-            query_cache.Query(gpu_addr, VideoCore::QueryType::SamplesPassed, {gpu.GetTicks()});
-        } else {
-            query_cache.Query(gpu_addr, VideoCore::QueryType::SamplesPassed, std::nullopt);
-        }
-        return;
+    const auto query_cache_type = MaxwellToVideoCoreQuery(type);
+    if (!query_cache_type.has_value()) {
+        return QueryFallback(gpu_addr, type, flags, payload, subreport);
     }
+    const bool has_timeout = True(flags & VideoCommon::QueryPropertiesFlags::HasTimeout);
+    const auto timestamp = has_timeout ? std::optional<u64>{gpu.GetTicks()} : std::nullopt;
+    query_cache.Query(gpu_addr, *query_cache_type, timestamp);
+}
+
+void RasterizerOpenGL::QueryFallback(GPUVAddr gpu_addr, VideoCommon::QueryType type,
+                                     VideoCommon::QueryPropertiesFlags flags, u32 payload,
+                                     u32 subreport) {
     if (type != VideoCommon::QueryType::Payload) {
         payload = 1u;
     }

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -225,6 +225,9 @@ private:
     /// End a transform feedback
     void EndTransformFeedback();
 
+    void QueryFallback(GPUVAddr gpu_addr, VideoCommon::QueryType type,
+                       VideoCommon::QueryPropertiesFlags flags, u32 payload, u32 subreport);
+
     Tegra::GPU& gpu;
 
     const Device& device;

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -51,7 +51,7 @@ using VideoCommon::LoadPipelines;
 using VideoCommon::SerializePipeline;
 using Context = ShaderContext::Context;
 
-constexpr u32 CACHE_VERSION = 9;
+constexpr u32 CACHE_VERSION = 10;
 
 template <typename Container>
 auto MakeSpan(Container& container) {
@@ -245,7 +245,6 @@ ShaderCache::ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindo
           .min_ssbo_alignment = static_cast<u32>(device.GetShaderStorageBufferAlignment()),
           .support_geometry_shader_passthrough = device.HasGeometryShaderPassthrough(),
           .support_conditional_barrier = device.SupportsConditionalBarriers(),
-          .support_ufloat_write_as_uint = true,
       } {
     if (use_asynchronous_shaders) {
         workers = CreateWorkers();

src/video_core/renderer_vulkan/vk_buffer_cache.cpp

@@ -78,8 +78,15 @@ vk::Buffer CreateBuffer(const Device& device, const MemoryAllocator& memory_allo
 }
 } // Anonymous namespace
 
-Buffer::Buffer(BufferCacheRuntime&, VideoCommon::NullBufferParams null_params)
-    : VideoCommon::BufferBase<VideoCore::RasterizerInterface>(null_params), tracker{4096} {}
+Buffer::Buffer(BufferCacheRuntime& runtime, VideoCommon::NullBufferParams null_params)
+    : VideoCommon::BufferBase<VideoCore::RasterizerInterface>(null_params), tracker{4096} {
+    if (runtime.device.HasNullDescriptor()) {
+        return;
+    }
+    device = &runtime.device;
+    buffer = runtime.CreateNullBuffer();
+    is_null = true;
+}
 
 Buffer::Buffer(BufferCacheRuntime& runtime, VideoCore::RasterizerInterface& rasterizer_,
                VAddr cpu_addr_, u64 size_bytes_)
@@ -93,8 +100,12 @@ Buffer::Buffer(BufferCacheRuntime& runtime, VideoCore::RasterizerInterface& rast
 
 VkBufferView Buffer::View(u32 offset, u32 size, VideoCore::Surface::PixelFormat format) {
     if (!device) {
-        // Null buffer, return a null descriptor
+        // Null buffer supported, return a null descriptor
         return VK_NULL_HANDLE;
+    } else if (is_null) {
+        // Null buffer not supported, adjust offset and size
+        offset = 0;
+        size = 0;
     }
     const auto it{std::ranges::find_if(views, [offset, size, format](const BufferView& view) {
         return offset == view.offset && size == view.size && format == view.format;
@@ -627,9 +638,12 @@ void BufferCacheRuntime::BindTransformFeedbackBuffers(VideoCommon::HostBindings<
 }
 
 void BufferCacheRuntime::ReserveNullBuffer() {
-    if (null_buffer) {
-        return;
+    if (!null_buffer) {
+        null_buffer = CreateNullBuffer();
     }
+}
+
+vk::Buffer BufferCacheRuntime::CreateNullBuffer() {
     VkBufferCreateInfo create_info{
         .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
         .pNext = nullptr,
@@ -644,15 +658,17 @@ void BufferCacheRuntime::ReserveNullBuffer() {
     if (device.IsExtTransformFeedbackSupported()) {
         create_info.usage |= VK_BUFFER_USAGE_TRANSFORM_FEEDBACK_BUFFER_BIT_EXT;
     }
-    null_buffer = memory_allocator.CreateBuffer(create_info, MemoryUsage::DeviceLocal);
+    vk::Buffer ret = memory_allocator.CreateBuffer(create_info, MemoryUsage::DeviceLocal);
     if (device.HasDebuggingToolAttached()) {
-        null_buffer.SetObjectNameEXT("Null buffer");
+        ret.SetObjectNameEXT("Null buffer");
     }
 
     scheduler.RequestOutsideRenderPassOperationContext();
-    scheduler.Record([buffer = *null_buffer](vk::CommandBuffer cmdbuf) {
+    scheduler.Record([buffer = *ret](vk::CommandBuffer cmdbuf) {
         cmdbuf.FillBuffer(buffer, 0, VK_WHOLE_SIZE, 0);
     });
+
+    return ret;
 }
 
 } // namespace Vulkan

src/video_core/renderer_vulkan/vk_buffer_cache.h

@@ -63,6 +63,7 @@ private:
     vk::Buffer buffer;
     std::vector<BufferView> views;
     VideoCommon::UsageTracker tracker;
+    bool is_null{};
 };
 
 class QuadArrayIndexBuffer;
@@ -151,6 +152,7 @@ private:
     }
 
     void ReserveNullBuffer();
+    vk::Buffer CreateNullBuffer();
 
     const Device& device;
     MemoryAllocator& memory_allocator;

src/video_core/renderer_vulkan/vk_pipeline_cache.cpp

@@ -54,7 +54,7 @@ using VideoCommon::FileEnvironment;
 using VideoCommon::GenericEnvironment;
 using VideoCommon::GraphicsEnvironment;
 
-constexpr u32 CACHE_VERSION = 10;
+constexpr u32 CACHE_VERSION = 11;
 constexpr std::array<char, 8> VULKAN_CACHE_MAGIC_NUMBER{'y', 'u', 'z', 'u', 'v', 'k', 'c', 'h'};
 
 template <typename Container>
@@ -388,8 +388,6 @@ PipelineCache::PipelineCache(RasterizerVulkan& rasterizer_, const Device& device
         .min_ssbo_alignment = static_cast<u32>(device.GetStorageBufferAlignment()),
         .support_geometry_shader_passthrough = device.IsNvGeometryShaderPassthroughSupported(),
         .support_conditional_barrier = device.SupportsConditionalBarriers(),
-        .support_ufloat_write_as_uint =
-            driver_id != VK_DRIVER_ID_QUALCOMM_PROPRIETARY && driver_id != VK_DRIVER_ID_MESA_TURNIP,
     };
 
     if (device.GetMaxVertexInputAttributes() < Maxwell::NumVertexAttributes) {

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -485,6 +485,10 @@ void RasterizerVulkan::DispatchCompute() {
 }
 
 void RasterizerVulkan::ResetCounter(VideoCommon::QueryType type) {
+    if (type != VideoCommon::QueryType::ZPassPixelCount64) {
+        LOG_DEBUG(Render_Vulkan, "Unimplemented counter reset={}", type);
+        return;
+    }
     query_cache.CounterReset(type);
 }
 

src/video_core/shader_environment.cpp

@@ -62,25 +62,9 @@ static Shader::TextureType ConvertTextureType(const Tegra::Texture::TICEntry& en
 }
 
 static Shader::TexturePixelFormat ConvertTexturePixelFormat(const Tegra::Texture::TICEntry& entry) {
-    switch (PixelFormatFromTextureInfo(entry.format, entry.r_type, entry.g_type, entry.b_type,
-                                       entry.a_type, entry.srgb_conversion)) {
-    case VideoCore::Surface::PixelFormat::A8B8G8R8_SNORM:
-        return Shader::TexturePixelFormat::A8B8G8R8_SNORM;
-    case VideoCore::Surface::PixelFormat::R8_SNORM:
-        return Shader::TexturePixelFormat::R8_SNORM;
-    case VideoCore::Surface::PixelFormat::R8G8_SNORM:
-        return Shader::TexturePixelFormat::R8G8_SNORM;
-    case VideoCore::Surface::PixelFormat::R16G16B16A16_SNORM:
-        return Shader::TexturePixelFormat::R16G16B16A16_SNORM;
-    case VideoCore::Surface::PixelFormat::R16G16_SNORM:
-        return Shader::TexturePixelFormat::R16G16_SNORM;
-    case VideoCore::Surface::PixelFormat::R16_SNORM:
-        return Shader::TexturePixelFormat::R16_SNORM;
-    case VideoCore::Surface::PixelFormat::B10G11R11_FLOAT:
-        return Shader::TexturePixelFormat::B10G11R11_FLOAT;
-    default:
-        return Shader::TexturePixelFormat::OTHER;
-    }
+    return static_cast<Shader::TexturePixelFormat>(
+        PixelFormatFromTextureInfo(entry.format, entry.r_type, entry.g_type, entry.b_type,
+                                   entry.a_type, entry.srgb_conversion));
 }
 
 static std::string_view StageToPrefix(Shader::Stage stage) {
@@ -400,6 +384,11 @@ Shader::TexturePixelFormat GraphicsEnvironment::ReadTexturePixelFormat(u32 handl
     return result;
 }
 
+bool GraphicsEnvironment::IsTexturePixelFormatInteger(u32 handle) {
+    return VideoCore::Surface::IsPixelFormatInteger(
+        static_cast<VideoCore::Surface::PixelFormat>(ReadTexturePixelFormat(handle)));
+}
+
 u32 GraphicsEnvironment::ReadViewportTransformState() {
     const auto& regs{maxwell3d->regs};
     viewport_transform_state = regs.viewport_scale_offset_enabled;
@@ -450,6 +439,11 @@ Shader::TexturePixelFormat ComputeEnvironment::ReadTexturePixelFormat(u32 handle
     return result;
 }
 
+bool ComputeEnvironment::IsTexturePixelFormatInteger(u32 handle) {
+    return VideoCore::Surface::IsPixelFormatInteger(
+        static_cast<VideoCore::Surface::PixelFormat>(ReadTexturePixelFormat(handle)));
+}
+
 u32 ComputeEnvironment::ReadViewportTransformState() {
     return viewport_transform_state;
 }
@@ -553,6 +547,11 @@ Shader::TexturePixelFormat FileEnvironment::ReadTexturePixelFormat(u32 handle) {
     return it->second;
 }
 
+bool FileEnvironment::IsTexturePixelFormatInteger(u32 handle) {
+    return VideoCore::Surface::IsPixelFormatInteger(
+        static_cast<VideoCore::Surface::PixelFormat>(ReadTexturePixelFormat(handle)));
+}
+
 u32 FileEnvironment::ReadViewportTransformState() {
     return viewport_transform_state;
 }

src/video_core/shader_environment.h

@@ -115,6 +115,8 @@ public:
 
     Shader::TexturePixelFormat ReadTexturePixelFormat(u32 handle) override;
 
+    bool IsTexturePixelFormatInteger(u32 handle) override;
+
     u32 ReadViewportTransformState() override;
 
     std::optional<Shader::ReplaceConstant> GetReplaceConstBuffer(u32 bank, u32 offset) override;
@@ -139,6 +141,8 @@ public:
 
     Shader::TexturePixelFormat ReadTexturePixelFormat(u32 handle) override;
 
+    bool IsTexturePixelFormatInteger(u32 handle) override;
+
     u32 ReadViewportTransformState() override;
 
     std::optional<Shader::ReplaceConstant> GetReplaceConstBuffer(
@@ -171,6 +175,8 @@ public:
 
     [[nodiscard]] Shader::TexturePixelFormat ReadTexturePixelFormat(u32 handle) override;
 
+    [[nodiscard]] bool IsTexturePixelFormatInteger(u32 handle) override;
+
     [[nodiscard]] u32 ReadViewportTransformState() override;
 
     [[nodiscard]] u32 LocalMemorySize() const override;

src/video_core/vulkan_common/vulkan_wrapper.cpp

@@ -246,7 +246,9 @@ void SetObjectName(const DeviceDispatch* dld, VkDevice device, T handle, VkObjec
         .objectHandle = reinterpret_cast<u64>(handle),
         .pObjectName = name,
     };
-    Check(dld->vkSetDebugUtilsObjectNameEXT(device, &name_info));
+    if (dld->vkSetDebugUtilsObjectNameEXT) {
+        Check(dld->vkSetDebugUtilsObjectNameEXT(device, &name_info));
+    }
 }
 
 } // Anonymous namespace