Merge pull request #3693 from ReinUsesLisp/clean-samplers
shader/texture: Support multiple unknown sampler properties
This commit is contained in:
commit
e6b4311178
13 changed files with 260 additions and 337 deletions
|
@ -59,14 +59,12 @@ constexpr std::size_t NumSupportedVertexAttributes = 16;
|
||||||
template <typename Engine, typename Entry>
|
template <typename Engine, typename Entry>
|
||||||
Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry,
|
Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry,
|
||||||
ShaderType shader_type, std::size_t index = 0) {
|
ShaderType shader_type, std::size_t index = 0) {
|
||||||
if (entry.IsBindless()) {
|
if (entry.is_bindless) {
|
||||||
const Tegra::Texture::TextureHandle tex_handle =
|
const auto tex_handle = engine.AccessConstBuffer32(shader_type, entry.buffer, entry.offset);
|
||||||
engine.AccessConstBuffer32(shader_type, entry.GetBuffer(), entry.GetOffset());
|
|
||||||
return engine.GetTextureInfo(tex_handle);
|
return engine.GetTextureInfo(tex_handle);
|
||||||
}
|
}
|
||||||
const auto& gpu_profile = engine.AccessGuestDriverProfile();
|
const auto& gpu_profile = engine.AccessGuestDriverProfile();
|
||||||
const u32 offset =
|
const u32 offset = entry.offset + static_cast<u32>(index * gpu_profile.GetTextureHandlerSize());
|
||||||
entry.GetOffset() + static_cast<u32>(index * gpu_profile.GetTextureHandlerSize());
|
|
||||||
if constexpr (std::is_same_v<Engine, Tegra::Engines::Maxwell3D>) {
|
if constexpr (std::is_same_v<Engine, Tegra::Engines::Maxwell3D>) {
|
||||||
return engine.GetStageTexture(shader_type, offset);
|
return engine.GetStageTexture(shader_type, offset);
|
||||||
} else {
|
} else {
|
||||||
|
@ -856,9 +854,9 @@ void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, const Shad
|
||||||
|
|
||||||
u32 binding = device.GetBaseBindings(stage_index).shader_storage_buffer;
|
u32 binding = device.GetBaseBindings(stage_index).shader_storage_buffer;
|
||||||
for (const auto& entry : shader->GetEntries().global_memory_entries) {
|
for (const auto& entry : shader->GetEntries().global_memory_entries) {
|
||||||
const auto addr{cbufs.const_buffers[entry.GetCbufIndex()].address + entry.GetCbufOffset()};
|
const GPUVAddr addr{cbufs.const_buffers[entry.cbuf_index].address + entry.cbuf_offset};
|
||||||
const auto gpu_addr{memory_manager.Read<u64>(addr)};
|
const GPUVAddr gpu_addr{memory_manager.Read<u64>(addr)};
|
||||||
const auto size{memory_manager.Read<u32>(addr + 8)};
|
const u32 size{memory_manager.Read<u32>(addr + 8)};
|
||||||
SetupGlobalMemory(binding++, entry, gpu_addr, size);
|
SetupGlobalMemory(binding++, entry, gpu_addr, size);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
@ -870,7 +868,7 @@ void RasterizerOpenGL::SetupComputeGlobalMemory(const Shader& kernel) {
|
||||||
|
|
||||||
u32 binding = 0;
|
u32 binding = 0;
|
||||||
for (const auto& entry : kernel->GetEntries().global_memory_entries) {
|
for (const auto& entry : kernel->GetEntries().global_memory_entries) {
|
||||||
const auto addr{cbufs[entry.GetCbufIndex()].Address() + entry.GetCbufOffset()};
|
const auto addr{cbufs[entry.cbuf_index].Address() + entry.cbuf_offset};
|
||||||
const auto gpu_addr{memory_manager.Read<u64>(addr)};
|
const auto gpu_addr{memory_manager.Read<u64>(addr)};
|
||||||
const auto size{memory_manager.Read<u32>(addr + 8)};
|
const auto size{memory_manager.Read<u32>(addr + 8)};
|
||||||
SetupGlobalMemory(binding++, entry, gpu_addr, size);
|
SetupGlobalMemory(binding++, entry, gpu_addr, size);
|
||||||
|
@ -881,7 +879,7 @@ void RasterizerOpenGL::SetupGlobalMemory(u32 binding, const GlobalMemoryEntry& e
|
||||||
GPUVAddr gpu_addr, std::size_t size) {
|
GPUVAddr gpu_addr, std::size_t size) {
|
||||||
const auto alignment{device.GetShaderStorageBufferAlignment()};
|
const auto alignment{device.GetShaderStorageBufferAlignment()};
|
||||||
const auto [ssbo, buffer_offset] =
|
const auto [ssbo, buffer_offset] =
|
||||||
buffer_cache.UploadMemory(gpu_addr, size, alignment, entry.IsWritten());
|
buffer_cache.UploadMemory(gpu_addr, size, alignment, entry.is_written);
|
||||||
glBindBufferRange(GL_SHADER_STORAGE_BUFFER, binding, ssbo, buffer_offset,
|
glBindBufferRange(GL_SHADER_STORAGE_BUFFER, binding, ssbo, buffer_offset,
|
||||||
static_cast<GLsizeiptr>(size));
|
static_cast<GLsizeiptr>(size));
|
||||||
}
|
}
|
||||||
|
@ -892,7 +890,7 @@ void RasterizerOpenGL::SetupDrawTextures(std::size_t stage_index, const Shader&
|
||||||
u32 binding = device.GetBaseBindings(stage_index).sampler;
|
u32 binding = device.GetBaseBindings(stage_index).sampler;
|
||||||
for (const auto& entry : shader->GetEntries().samplers) {
|
for (const auto& entry : shader->GetEntries().samplers) {
|
||||||
const auto shader_type = static_cast<ShaderType>(stage_index);
|
const auto shader_type = static_cast<ShaderType>(stage_index);
|
||||||
for (std::size_t i = 0; i < entry.Size(); ++i) {
|
for (std::size_t i = 0; i < entry.size; ++i) {
|
||||||
const auto texture = GetTextureInfo(maxwell3d, entry, shader_type, i);
|
const auto texture = GetTextureInfo(maxwell3d, entry, shader_type, i);
|
||||||
SetupTexture(binding++, texture, entry);
|
SetupTexture(binding++, texture, entry);
|
||||||
}
|
}
|
||||||
|
@ -904,7 +902,7 @@ void RasterizerOpenGL::SetupComputeTextures(const Shader& kernel) {
|
||||||
const auto& compute = system.GPU().KeplerCompute();
|
const auto& compute = system.GPU().KeplerCompute();
|
||||||
u32 binding = 0;
|
u32 binding = 0;
|
||||||
for (const auto& entry : kernel->GetEntries().samplers) {
|
for (const auto& entry : kernel->GetEntries().samplers) {
|
||||||
for (std::size_t i = 0; i < entry.Size(); ++i) {
|
for (std::size_t i = 0; i < entry.size; ++i) {
|
||||||
const auto texture = GetTextureInfo(compute, entry, ShaderType::Compute, i);
|
const auto texture = GetTextureInfo(compute, entry, ShaderType::Compute, i);
|
||||||
SetupTexture(binding++, texture, entry);
|
SetupTexture(binding++, texture, entry);
|
||||||
}
|
}
|
||||||
|
@ -961,7 +959,7 @@ void RasterizerOpenGL::SetupImage(u32 binding, const Tegra::Texture::TICEntry& t
|
||||||
if (!tic.IsBuffer()) {
|
if (!tic.IsBuffer()) {
|
||||||
view->ApplySwizzle(tic.x_source, tic.y_source, tic.z_source, tic.w_source);
|
view->ApplySwizzle(tic.x_source, tic.y_source, tic.z_source, tic.w_source);
|
||||||
}
|
}
|
||||||
if (entry.IsWritten()) {
|
if (entry.is_written) {
|
||||||
view->MarkAsModified(texture_cache.Tick());
|
view->MarkAsModified(texture_cache.Tick());
|
||||||
}
|
}
|
||||||
glBindImageTexture(binding, view->GetTexture(), 0, GL_TRUE, 0, GL_READ_WRITE,
|
glBindImageTexture(binding, view->GetTexture(), 0, GL_TRUE, 0, GL_READ_WRITE,
|
||||||
|
|
|
@ -870,13 +870,13 @@ private:
|
||||||
for (const auto& sampler : ir.GetSamplers()) {
|
for (const auto& sampler : ir.GetSamplers()) {
|
||||||
const std::string name = GetSampler(sampler);
|
const std::string name = GetSampler(sampler);
|
||||||
const std::string description = fmt::format("layout (binding = {}) uniform", binding);
|
const std::string description = fmt::format("layout (binding = {}) uniform", binding);
|
||||||
binding += sampler.IsIndexed() ? sampler.Size() : 1;
|
binding += sampler.is_indexed ? sampler.size : 1;
|
||||||
|
|
||||||
std::string sampler_type = [&]() {
|
std::string sampler_type = [&]() {
|
||||||
if (sampler.IsBuffer()) {
|
if (sampler.is_buffer) {
|
||||||
return "samplerBuffer";
|
return "samplerBuffer";
|
||||||
}
|
}
|
||||||
switch (sampler.GetType()) {
|
switch (sampler.type) {
|
||||||
case Tegra::Shader::TextureType::Texture1D:
|
case Tegra::Shader::TextureType::Texture1D:
|
||||||
return "sampler1D";
|
return "sampler1D";
|
||||||
case Tegra::Shader::TextureType::Texture2D:
|
case Tegra::Shader::TextureType::Texture2D:
|
||||||
|
@ -890,17 +890,17 @@ private:
|
||||||
return "sampler2D";
|
return "sampler2D";
|
||||||
}
|
}
|
||||||
}();
|
}();
|
||||||
if (sampler.IsArray()) {
|
if (sampler.is_array) {
|
||||||
sampler_type += "Array";
|
sampler_type += "Array";
|
||||||
}
|
}
|
||||||
if (sampler.IsShadow()) {
|
if (sampler.is_shadow) {
|
||||||
sampler_type += "Shadow";
|
sampler_type += "Shadow";
|
||||||
}
|
}
|
||||||
|
|
||||||
if (!sampler.IsIndexed()) {
|
if (!sampler.is_indexed) {
|
||||||
code.AddLine("{} {} {};", description, sampler_type, name);
|
code.AddLine("{} {} {};", description, sampler_type, name);
|
||||||
} else {
|
} else {
|
||||||
code.AddLine("{} {} {}[{}];", description, sampler_type, name, sampler.Size());
|
code.AddLine("{} {} {}[{}];", description, sampler_type, name, sampler.size);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
if (!ir.GetSamplers().empty()) {
|
if (!ir.GetSamplers().empty()) {
|
||||||
|
@ -946,14 +946,14 @@ private:
|
||||||
u32 binding = device.GetBaseBindings(stage).image;
|
u32 binding = device.GetBaseBindings(stage).image;
|
||||||
for (const auto& image : ir.GetImages()) {
|
for (const auto& image : ir.GetImages()) {
|
||||||
std::string qualifier = "coherent volatile";
|
std::string qualifier = "coherent volatile";
|
||||||
if (image.IsRead() && !image.IsWritten()) {
|
if (image.is_read && !image.is_written) {
|
||||||
qualifier += " readonly";
|
qualifier += " readonly";
|
||||||
} else if (image.IsWritten() && !image.IsRead()) {
|
} else if (image.is_written && !image.is_read) {
|
||||||
qualifier += " writeonly";
|
qualifier += " writeonly";
|
||||||
}
|
}
|
||||||
|
|
||||||
const char* format = image.IsAtomic() ? "r32ui, " : "";
|
const char* format = image.is_atomic ? "r32ui, " : "";
|
||||||
const char* type_declaration = GetImageTypeDeclaration(image.GetType());
|
const char* type_declaration = GetImageTypeDeclaration(image.type);
|
||||||
code.AddLine("layout ({}binding = {}) {} uniform uimage{} {};", format, binding++,
|
code.AddLine("layout ({}binding = {}) {} uniform uimage{} {};", format, binding++,
|
||||||
qualifier, type_declaration, GetImage(image));
|
qualifier, type_declaration, GetImage(image));
|
||||||
}
|
}
|
||||||
|
@ -1337,8 +1337,8 @@ private:
|
||||||
ASSERT(meta);
|
ASSERT(meta);
|
||||||
|
|
||||||
const std::size_t count = operation.GetOperandsCount();
|
const std::size_t count = operation.GetOperandsCount();
|
||||||
const bool has_array = meta->sampler.IsArray();
|
const bool has_array = meta->sampler.is_array;
|
||||||
const bool has_shadow = meta->sampler.IsShadow();
|
const bool has_shadow = meta->sampler.is_shadow;
|
||||||
|
|
||||||
std::string expr = "texture" + function_suffix;
|
std::string expr = "texture" + function_suffix;
|
||||||
if (!meta->aoffi.empty()) {
|
if (!meta->aoffi.empty()) {
|
||||||
|
@ -1346,7 +1346,7 @@ private:
|
||||||
} else if (!meta->ptp.empty()) {
|
} else if (!meta->ptp.empty()) {
|
||||||
expr += "Offsets";
|
expr += "Offsets";
|
||||||
}
|
}
|
||||||
if (!meta->sampler.IsIndexed()) {
|
if (!meta->sampler.is_indexed) {
|
||||||
expr += '(' + GetSampler(meta->sampler) + ", ";
|
expr += '(' + GetSampler(meta->sampler) + ", ";
|
||||||
} else {
|
} else {
|
||||||
expr += '(' + GetSampler(meta->sampler) + '[' + Visit(meta->index).AsUint() + "], ";
|
expr += '(' + GetSampler(meta->sampler) + '[' + Visit(meta->index).AsUint() + "], ";
|
||||||
|
@ -1982,7 +1982,7 @@ private:
|
||||||
|
|
||||||
std::string expr = GenerateTexture(
|
std::string expr = GenerateTexture(
|
||||||
operation, "", {TextureOffset{}, TextureArgument{Type::Float, meta->bias}});
|
operation, "", {TextureOffset{}, TextureArgument{Type::Float, meta->bias}});
|
||||||
if (meta->sampler.IsShadow()) {
|
if (meta->sampler.is_shadow) {
|
||||||
expr = "vec4(" + expr + ')';
|
expr = "vec4(" + expr + ')';
|
||||||
}
|
}
|
||||||
return {expr + GetSwizzle(meta->element), Type::Float};
|
return {expr + GetSwizzle(meta->element), Type::Float};
|
||||||
|
@ -1994,7 +1994,7 @@ private:
|
||||||
|
|
||||||
std::string expr = GenerateTexture(
|
std::string expr = GenerateTexture(
|
||||||
operation, "Lod", {TextureArgument{Type::Float, meta->lod}, TextureOffset{}});
|
operation, "Lod", {TextureArgument{Type::Float, meta->lod}, TextureOffset{}});
|
||||||
if (meta->sampler.IsShadow()) {
|
if (meta->sampler.is_shadow) {
|
||||||
expr = "vec4(" + expr + ')';
|
expr = "vec4(" + expr + ')';
|
||||||
}
|
}
|
||||||
return {expr + GetSwizzle(meta->element), Type::Float};
|
return {expr + GetSwizzle(meta->element), Type::Float};
|
||||||
|
@ -2003,11 +2003,11 @@ private:
|
||||||
Expression TextureGather(Operation operation) {
|
Expression TextureGather(Operation operation) {
|
||||||
const auto& meta = std::get<MetaTexture>(operation.GetMeta());
|
const auto& meta = std::get<MetaTexture>(operation.GetMeta());
|
||||||
|
|
||||||
const auto type = meta.sampler.IsShadow() ? Type::Float : Type::Int;
|
const auto type = meta.sampler.is_shadow ? Type::Float : Type::Int;
|
||||||
const bool separate_dc = meta.sampler.IsShadow();
|
const bool separate_dc = meta.sampler.is_shadow;
|
||||||
|
|
||||||
std::vector<TextureIR> ir;
|
std::vector<TextureIR> ir;
|
||||||
if (meta.sampler.IsShadow()) {
|
if (meta.sampler.is_shadow) {
|
||||||
ir = {TextureOffset{}};
|
ir = {TextureOffset{}};
|
||||||
} else {
|
} else {
|
||||||
ir = {TextureOffset{}, TextureArgument{type, meta.component}};
|
ir = {TextureOffset{}, TextureArgument{type, meta.component}};
|
||||||
|
@ -2052,7 +2052,7 @@ private:
|
||||||
constexpr std::array constructors = {"int", "ivec2", "ivec3", "ivec4"};
|
constexpr std::array constructors = {"int", "ivec2", "ivec3", "ivec4"};
|
||||||
const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
|
const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
|
||||||
ASSERT(meta);
|
ASSERT(meta);
|
||||||
UNIMPLEMENTED_IF(meta->sampler.IsArray());
|
UNIMPLEMENTED_IF(meta->sampler.is_array);
|
||||||
const std::size_t count = operation.GetOperandsCount();
|
const std::size_t count = operation.GetOperandsCount();
|
||||||
|
|
||||||
std::string expr = "texelFetch(";
|
std::string expr = "texelFetch(";
|
||||||
|
@ -2073,7 +2073,7 @@ private:
|
||||||
}
|
}
|
||||||
expr += ')';
|
expr += ')';
|
||||||
|
|
||||||
if (meta->lod && !meta->sampler.IsBuffer()) {
|
if (meta->lod && !meta->sampler.is_buffer) {
|
||||||
expr += ", ";
|
expr += ", ";
|
||||||
expr += Visit(meta->lod).AsInt();
|
expr += Visit(meta->lod).AsInt();
|
||||||
}
|
}
|
||||||
|
@ -2084,12 +2084,10 @@ private:
|
||||||
}
|
}
|
||||||
|
|
||||||
Expression TextureGradient(Operation operation) {
|
Expression TextureGradient(Operation operation) {
|
||||||
const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
|
const auto& meta = std::get<MetaTexture>(operation.GetMeta());
|
||||||
ASSERT(meta);
|
|
||||||
|
|
||||||
std::string expr =
|
std::string expr =
|
||||||
GenerateTexture(operation, "Grad", {TextureDerivates{}, TextureOffset{}});
|
GenerateTexture(operation, "Grad", {TextureDerivates{}, TextureOffset{}});
|
||||||
return {std::move(expr) + GetSwizzle(meta->element), Type::Float};
|
return {std::move(expr) + GetSwizzle(meta.element), Type::Float};
|
||||||
}
|
}
|
||||||
|
|
||||||
Expression ImageLoad(Operation operation) {
|
Expression ImageLoad(Operation operation) {
|
||||||
|
@ -2608,11 +2606,11 @@ private:
|
||||||
}
|
}
|
||||||
|
|
||||||
std::string GetSampler(const Sampler& sampler) const {
|
std::string GetSampler(const Sampler& sampler) const {
|
||||||
return AppendSuffix(static_cast<u32>(sampler.GetIndex()), "sampler");
|
return AppendSuffix(sampler.index, "sampler");
|
||||||
}
|
}
|
||||||
|
|
||||||
std::string GetImage(const Image& image) const {
|
std::string GetImage(const Image& image) const {
|
||||||
return AppendSuffix(static_cast<u32>(image.GetIndex()), "image");
|
return AppendSuffix(image.index, "image");
|
||||||
}
|
}
|
||||||
|
|
||||||
std::string AppendSuffix(u32 index, std::string_view name) const {
|
std::string AppendSuffix(u32 index, std::string_view name) const {
|
||||||
|
|
|
@ -33,36 +33,19 @@ public:
|
||||||
}
|
}
|
||||||
|
|
||||||
private:
|
private:
|
||||||
u32 index{};
|
u32 index = 0;
|
||||||
};
|
};
|
||||||
|
|
||||||
class GlobalMemoryEntry {
|
struct GlobalMemoryEntry {
|
||||||
public:
|
constexpr explicit GlobalMemoryEntry(u32 cbuf_index, u32 cbuf_offset, bool is_read,
|
||||||
explicit GlobalMemoryEntry(u32 cbuf_index, u32 cbuf_offset, bool is_read, bool is_written)
|
bool is_written)
|
||||||
: cbuf_index{cbuf_index}, cbuf_offset{cbuf_offset}, is_read{is_read}, is_written{
|
: cbuf_index{cbuf_index}, cbuf_offset{cbuf_offset}, is_read{is_read}, is_written{
|
||||||
is_written} {}
|
is_written} {}
|
||||||
|
|
||||||
u32 GetCbufIndex() const {
|
u32 cbuf_index = 0;
|
||||||
return cbuf_index;
|
u32 cbuf_offset = 0;
|
||||||
}
|
bool is_read = false;
|
||||||
|
bool is_written = false;
|
||||||
u32 GetCbufOffset() const {
|
|
||||||
return cbuf_offset;
|
|
||||||
}
|
|
||||||
|
|
||||||
bool IsRead() const {
|
|
||||||
return is_read;
|
|
||||||
}
|
|
||||||
|
|
||||||
bool IsWritten() const {
|
|
||||||
return is_written;
|
|
||||||
}
|
|
||||||
|
|
||||||
private:
|
|
||||||
u32 cbuf_index{};
|
|
||||||
u32 cbuf_offset{};
|
|
||||||
bool is_read{};
|
|
||||||
bool is_written{};
|
|
||||||
};
|
};
|
||||||
|
|
||||||
struct ShaderEntries {
|
struct ShaderEntries {
|
||||||
|
|
|
@ -85,7 +85,7 @@ void AddBindings(std::vector<VkDescriptorSetLayoutBinding>& bindings, u32& bindi
|
||||||
u32 count = 1;
|
u32 count = 1;
|
||||||
if constexpr (descriptor_type == VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER) {
|
if constexpr (descriptor_type == VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER) {
|
||||||
// Combined image samplers can be arrayed.
|
// Combined image samplers can be arrayed.
|
||||||
count = container[i].Size();
|
count = container[i].size;
|
||||||
}
|
}
|
||||||
VkDescriptorSetLayoutBinding& entry = bindings.emplace_back();
|
VkDescriptorSetLayoutBinding& entry = bindings.emplace_back();
|
||||||
entry.binding = binding++;
|
entry.binding = binding++;
|
||||||
|
@ -362,7 +362,7 @@ void AddEntry(std::vector<VkDescriptorUpdateTemplateEntry>& template_entries, u3
|
||||||
|
|
||||||
if constexpr (descriptor_type == COMBINED_IMAGE_SAMPLER) {
|
if constexpr (descriptor_type == COMBINED_IMAGE_SAMPLER) {
|
||||||
for (u32 i = 0; i < count; ++i) {
|
for (u32 i = 0; i < count; ++i) {
|
||||||
const u32 num_samplers = container[i].Size();
|
const u32 num_samplers = container[i].size;
|
||||||
VkDescriptorUpdateTemplateEntry& entry = template_entries.emplace_back();
|
VkDescriptorUpdateTemplateEntry& entry = template_entries.emplace_back();
|
||||||
entry.dstBinding = binding;
|
entry.dstBinding = binding;
|
||||||
entry.dstArrayElement = 0;
|
entry.dstArrayElement = 0;
|
||||||
|
|
|
@ -119,14 +119,13 @@ template <typename Engine, typename Entry>
|
||||||
Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry,
|
Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry,
|
||||||
std::size_t stage, std::size_t index = 0) {
|
std::size_t stage, std::size_t index = 0) {
|
||||||
const auto stage_type = static_cast<Tegra::Engines::ShaderType>(stage);
|
const auto stage_type = static_cast<Tegra::Engines::ShaderType>(stage);
|
||||||
if (entry.IsBindless()) {
|
if (entry.is_bindless) {
|
||||||
const Tegra::Texture::TextureHandle tex_handle =
|
const auto tex_handle = engine.AccessConstBuffer32(stage_type, entry.buffer, entry.offset);
|
||||||
engine.AccessConstBuffer32(stage_type, entry.GetBuffer(), entry.GetOffset());
|
|
||||||
return engine.GetTextureInfo(tex_handle);
|
return engine.GetTextureInfo(tex_handle);
|
||||||
}
|
}
|
||||||
const auto& gpu_profile = engine.AccessGuestDriverProfile();
|
const auto& gpu_profile = engine.AccessGuestDriverProfile();
|
||||||
const u32 entry_offset = static_cast<u32>(index * gpu_profile.GetTextureHandlerSize());
|
const u32 entry_offset = static_cast<u32>(index * gpu_profile.GetTextureHandlerSize());
|
||||||
const u32 offset = entry.GetOffset() + entry_offset;
|
const u32 offset = entry.offset + entry_offset;
|
||||||
if constexpr (std::is_same_v<Engine, Tegra::Engines::Maxwell3D>) {
|
if constexpr (std::is_same_v<Engine, Tegra::Engines::Maxwell3D>) {
|
||||||
return engine.GetStageTexture(stage_type, offset);
|
return engine.GetStageTexture(stage_type, offset);
|
||||||
} else {
|
} else {
|
||||||
|
@ -971,7 +970,7 @@ void RasterizerVulkan::SetupGraphicsTextures(const ShaderEntries& entries, std::
|
||||||
MICROPROFILE_SCOPE(Vulkan_Textures);
|
MICROPROFILE_SCOPE(Vulkan_Textures);
|
||||||
const auto& gpu = system.GPU().Maxwell3D();
|
const auto& gpu = system.GPU().Maxwell3D();
|
||||||
for (const auto& entry : entries.samplers) {
|
for (const auto& entry : entries.samplers) {
|
||||||
for (std::size_t i = 0; i < entry.Size(); ++i) {
|
for (std::size_t i = 0; i < entry.size; ++i) {
|
||||||
const auto texture = GetTextureInfo(gpu, entry, stage, i);
|
const auto texture = GetTextureInfo(gpu, entry, stage, i);
|
||||||
SetupTexture(texture, entry);
|
SetupTexture(texture, entry);
|
||||||
}
|
}
|
||||||
|
@ -1023,7 +1022,7 @@ void RasterizerVulkan::SetupComputeTextures(const ShaderEntries& entries) {
|
||||||
MICROPROFILE_SCOPE(Vulkan_Textures);
|
MICROPROFILE_SCOPE(Vulkan_Textures);
|
||||||
const auto& gpu = system.GPU().KeplerCompute();
|
const auto& gpu = system.GPU().KeplerCompute();
|
||||||
for (const auto& entry : entries.samplers) {
|
for (const auto& entry : entries.samplers) {
|
||||||
for (std::size_t i = 0; i < entry.Size(); ++i) {
|
for (std::size_t i = 0; i < entry.size; ++i) {
|
||||||
const auto texture = GetTextureInfo(gpu, entry, ComputeShaderIndex, i);
|
const auto texture = GetTextureInfo(gpu, entry, ComputeShaderIndex, i);
|
||||||
SetupTexture(texture, entry);
|
SetupTexture(texture, entry);
|
||||||
}
|
}
|
||||||
|
@ -1105,7 +1104,7 @@ void RasterizerVulkan::SetupTexture(const Tegra::Texture::FullTextureInfo& textu
|
||||||
void RasterizerVulkan::SetupImage(const Tegra::Texture::TICEntry& tic, const ImageEntry& entry) {
|
void RasterizerVulkan::SetupImage(const Tegra::Texture::TICEntry& tic, const ImageEntry& entry) {
|
||||||
auto view = texture_cache.GetImageSurface(tic, entry);
|
auto view = texture_cache.GetImageSurface(tic, entry);
|
||||||
|
|
||||||
if (entry.IsWritten()) {
|
if (entry.is_written) {
|
||||||
view->MarkAsModified(texture_cache.Tick());
|
view->MarkAsModified(texture_cache.Tick());
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
|
@ -103,8 +103,8 @@ struct GenericVaryingDescription {
|
||||||
};
|
};
|
||||||
|
|
||||||
spv::Dim GetSamplerDim(const Sampler& sampler) {
|
spv::Dim GetSamplerDim(const Sampler& sampler) {
|
||||||
ASSERT(!sampler.IsBuffer());
|
ASSERT(!sampler.is_buffer);
|
||||||
switch (sampler.GetType()) {
|
switch (sampler.type) {
|
||||||
case Tegra::Shader::TextureType::Texture1D:
|
case Tegra::Shader::TextureType::Texture1D:
|
||||||
return spv::Dim::Dim1D;
|
return spv::Dim::Dim1D;
|
||||||
case Tegra::Shader::TextureType::Texture2D:
|
case Tegra::Shader::TextureType::Texture2D:
|
||||||
|
@ -114,13 +114,13 @@ spv::Dim GetSamplerDim(const Sampler& sampler) {
|
||||||
case Tegra::Shader::TextureType::TextureCube:
|
case Tegra::Shader::TextureType::TextureCube:
|
||||||
return spv::Dim::Cube;
|
return spv::Dim::Cube;
|
||||||
default:
|
default:
|
||||||
UNIMPLEMENTED_MSG("Unimplemented sampler type={}", static_cast<u32>(sampler.GetType()));
|
UNIMPLEMENTED_MSG("Unimplemented sampler type={}", static_cast<int>(sampler.type));
|
||||||
return spv::Dim::Dim2D;
|
return spv::Dim::Dim2D;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
std::pair<spv::Dim, bool> GetImageDim(const Image& image) {
|
std::pair<spv::Dim, bool> GetImageDim(const Image& image) {
|
||||||
switch (image.GetType()) {
|
switch (image.type) {
|
||||||
case Tegra::Shader::ImageType::Texture1D:
|
case Tegra::Shader::ImageType::Texture1D:
|
||||||
return {spv::Dim::Dim1D, false};
|
return {spv::Dim::Dim1D, false};
|
||||||
case Tegra::Shader::ImageType::TextureBuffer:
|
case Tegra::Shader::ImageType::TextureBuffer:
|
||||||
|
@ -134,7 +134,7 @@ std::pair<spv::Dim, bool> GetImageDim(const Image& image) {
|
||||||
case Tegra::Shader::ImageType::Texture3D:
|
case Tegra::Shader::ImageType::Texture3D:
|
||||||
return {spv::Dim::Dim3D, false};
|
return {spv::Dim::Dim3D, false};
|
||||||
default:
|
default:
|
||||||
UNIMPLEMENTED_MSG("Unimplemented image type={}", static_cast<u32>(image.GetType()));
|
UNIMPLEMENTED_MSG("Unimplemented image type={}", static_cast<int>(image.type));
|
||||||
return {spv::Dim::Dim2D, false};
|
return {spv::Dim::Dim2D, false};
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
@ -879,11 +879,11 @@ private:
|
||||||
|
|
||||||
u32 DeclareTexelBuffers(u32 binding) {
|
u32 DeclareTexelBuffers(u32 binding) {
|
||||||
for (const auto& sampler : ir.GetSamplers()) {
|
for (const auto& sampler : ir.GetSamplers()) {
|
||||||
if (!sampler.IsBuffer()) {
|
if (!sampler.is_buffer) {
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
ASSERT(!sampler.IsArray());
|
ASSERT(!sampler.is_array);
|
||||||
ASSERT(!sampler.IsShadow());
|
ASSERT(!sampler.is_shadow);
|
||||||
|
|
||||||
constexpr auto dim = spv::Dim::Buffer;
|
constexpr auto dim = spv::Dim::Buffer;
|
||||||
constexpr int depth = 0;
|
constexpr int depth = 0;
|
||||||
|
@ -894,23 +894,23 @@ private:
|
||||||
const Id image_type = TypeImage(t_float, dim, depth, arrayed, ms, sampled, format);
|
const Id image_type = TypeImage(t_float, dim, depth, arrayed, ms, sampled, format);
|
||||||
const Id pointer_type = TypePointer(spv::StorageClass::UniformConstant, image_type);
|
const Id pointer_type = TypePointer(spv::StorageClass::UniformConstant, image_type);
|
||||||
const Id id = OpVariable(pointer_type, spv::StorageClass::UniformConstant);
|
const Id id = OpVariable(pointer_type, spv::StorageClass::UniformConstant);
|
||||||
AddGlobalVariable(Name(id, fmt::format("sampler_{}", sampler.GetIndex())));
|
AddGlobalVariable(Name(id, fmt::format("sampler_{}", sampler.index)));
|
||||||
Decorate(id, spv::Decoration::Binding, binding++);
|
Decorate(id, spv::Decoration::Binding, binding++);
|
||||||
Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET);
|
Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET);
|
||||||
|
|
||||||
texel_buffers.emplace(sampler.GetIndex(), TexelBuffer{image_type, id});
|
texel_buffers.emplace(sampler.index, TexelBuffer{image_type, id});
|
||||||
}
|
}
|
||||||
return binding;
|
return binding;
|
||||||
}
|
}
|
||||||
|
|
||||||
u32 DeclareSamplers(u32 binding) {
|
u32 DeclareSamplers(u32 binding) {
|
||||||
for (const auto& sampler : ir.GetSamplers()) {
|
for (const auto& sampler : ir.GetSamplers()) {
|
||||||
if (sampler.IsBuffer()) {
|
if (sampler.is_buffer) {
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
const auto dim = GetSamplerDim(sampler);
|
const auto dim = GetSamplerDim(sampler);
|
||||||
const int depth = sampler.IsShadow() ? 1 : 0;
|
const int depth = sampler.is_shadow ? 1 : 0;
|
||||||
const int arrayed = sampler.IsArray() ? 1 : 0;
|
const int arrayed = sampler.is_array ? 1 : 0;
|
||||||
constexpr bool ms = false;
|
constexpr bool ms = false;
|
||||||
constexpr int sampled = 1;
|
constexpr int sampled = 1;
|
||||||
constexpr auto format = spv::ImageFormat::Unknown;
|
constexpr auto format = spv::ImageFormat::Unknown;
|
||||||
|
@ -918,17 +918,17 @@ private:
|
||||||
const Id sampler_type = TypeSampledImage(image_type);
|
const Id sampler_type = TypeSampledImage(image_type);
|
||||||
const Id sampler_pointer_type =
|
const Id sampler_pointer_type =
|
||||||
TypePointer(spv::StorageClass::UniformConstant, sampler_type);
|
TypePointer(spv::StorageClass::UniformConstant, sampler_type);
|
||||||
const Id type = sampler.IsIndexed()
|
const Id type = sampler.is_indexed
|
||||||
? TypeArray(sampler_type, Constant(t_uint, sampler.Size()))
|
? TypeArray(sampler_type, Constant(t_uint, sampler.size))
|
||||||
: sampler_type;
|
: sampler_type;
|
||||||
const Id pointer_type = TypePointer(spv::StorageClass::UniformConstant, type);
|
const Id pointer_type = TypePointer(spv::StorageClass::UniformConstant, type);
|
||||||
const Id id = OpVariable(pointer_type, spv::StorageClass::UniformConstant);
|
const Id id = OpVariable(pointer_type, spv::StorageClass::UniformConstant);
|
||||||
AddGlobalVariable(Name(id, fmt::format("sampler_{}", sampler.GetIndex())));
|
AddGlobalVariable(Name(id, fmt::format("sampler_{}", sampler.index)));
|
||||||
Decorate(id, spv::Decoration::Binding, binding++);
|
Decorate(id, spv::Decoration::Binding, binding++);
|
||||||
Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET);
|
Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET);
|
||||||
|
|
||||||
sampled_images.emplace(sampler.GetIndex(), SampledImage{image_type, sampler_type,
|
sampled_images.emplace(
|
||||||
sampler_pointer_type, id});
|
sampler.index, SampledImage{image_type, sampler_type, sampler_pointer_type, id});
|
||||||
}
|
}
|
||||||
return binding;
|
return binding;
|
||||||
}
|
}
|
||||||
|
@ -943,17 +943,17 @@ private:
|
||||||
const Id image_type = TypeImage(t_uint, dim, depth, arrayed, ms, sampled, format, {});
|
const Id image_type = TypeImage(t_uint, dim, depth, arrayed, ms, sampled, format, {});
|
||||||
const Id pointer_type = TypePointer(spv::StorageClass::UniformConstant, image_type);
|
const Id pointer_type = TypePointer(spv::StorageClass::UniformConstant, image_type);
|
||||||
const Id id = OpVariable(pointer_type, spv::StorageClass::UniformConstant);
|
const Id id = OpVariable(pointer_type, spv::StorageClass::UniformConstant);
|
||||||
AddGlobalVariable(Name(id, fmt::format("image_{}", image.GetIndex())));
|
AddGlobalVariable(Name(id, fmt::format("image_{}", image.index)));
|
||||||
|
|
||||||
Decorate(id, spv::Decoration::Binding, binding++);
|
Decorate(id, spv::Decoration::Binding, binding++);
|
||||||
Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET);
|
Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET);
|
||||||
if (image.IsRead() && !image.IsWritten()) {
|
if (image.is_read && !image.is_written) {
|
||||||
Decorate(id, spv::Decoration::NonWritable);
|
Decorate(id, spv::Decoration::NonWritable);
|
||||||
} else if (image.IsWritten() && !image.IsRead()) {
|
} else if (image.is_written && !image.is_read) {
|
||||||
Decorate(id, spv::Decoration::NonReadable);
|
Decorate(id, spv::Decoration::NonReadable);
|
||||||
}
|
}
|
||||||
|
|
||||||
images.emplace(static_cast<u32>(image.GetIndex()), StorageImage{image_type, id});
|
images.emplace(image.index, StorageImage{image_type, id});
|
||||||
}
|
}
|
||||||
return binding;
|
return binding;
|
||||||
}
|
}
|
||||||
|
@ -1620,11 +1620,11 @@ private:
|
||||||
|
|
||||||
Id GetTextureSampler(Operation operation) {
|
Id GetTextureSampler(Operation operation) {
|
||||||
const auto& meta = std::get<MetaTexture>(operation.GetMeta());
|
const auto& meta = std::get<MetaTexture>(operation.GetMeta());
|
||||||
ASSERT(!meta.sampler.IsBuffer());
|
ASSERT(!meta.sampler.is_buffer);
|
||||||
|
|
||||||
const auto& entry = sampled_images.at(meta.sampler.GetIndex());
|
const auto& entry = sampled_images.at(meta.sampler.index);
|
||||||
Id sampler = entry.variable;
|
Id sampler = entry.variable;
|
||||||
if (meta.sampler.IsIndexed()) {
|
if (meta.sampler.is_indexed) {
|
||||||
const Id index = AsInt(Visit(meta.index));
|
const Id index = AsInt(Visit(meta.index));
|
||||||
sampler = OpAccessChain(entry.sampler_pointer_type, sampler, index);
|
sampler = OpAccessChain(entry.sampler_pointer_type, sampler, index);
|
||||||
}
|
}
|
||||||
|
@ -1633,8 +1633,8 @@ private:
|
||||||
|
|
||||||
Id GetTextureImage(Operation operation) {
|
Id GetTextureImage(Operation operation) {
|
||||||
const auto& meta = std::get<MetaTexture>(operation.GetMeta());
|
const auto& meta = std::get<MetaTexture>(operation.GetMeta());
|
||||||
const u32 index = meta.sampler.GetIndex();
|
const u32 index = meta.sampler.index;
|
||||||
if (meta.sampler.IsBuffer()) {
|
if (meta.sampler.is_buffer) {
|
||||||
const auto& entry = texel_buffers.at(index);
|
const auto& entry = texel_buffers.at(index);
|
||||||
return OpLoad(entry.image_type, entry.image);
|
return OpLoad(entry.image_type, entry.image);
|
||||||
} else {
|
} else {
|
||||||
|
@ -1645,7 +1645,7 @@ private:
|
||||||
|
|
||||||
Id GetImage(Operation operation) {
|
Id GetImage(Operation operation) {
|
||||||
const auto& meta = std::get<MetaImage>(operation.GetMeta());
|
const auto& meta = std::get<MetaImage>(operation.GetMeta());
|
||||||
const auto entry = images.at(meta.image.GetIndex());
|
const auto entry = images.at(meta.image.index);
|
||||||
return OpLoad(entry.image_type, entry.image);
|
return OpLoad(entry.image_type, entry.image);
|
||||||
}
|
}
|
||||||
|
|
||||||
|
@ -1661,7 +1661,7 @@ private:
|
||||||
}
|
}
|
||||||
if (const auto meta = std::get_if<MetaTexture>(&operation.GetMeta())) {
|
if (const auto meta = std::get_if<MetaTexture>(&operation.GetMeta())) {
|
||||||
// Add array coordinate for textures
|
// Add array coordinate for textures
|
||||||
if (meta->sampler.IsArray()) {
|
if (meta->sampler.is_array) {
|
||||||
Id array = AsInt(Visit(meta->array));
|
Id array = AsInt(Visit(meta->array));
|
||||||
if (type == Type::Float) {
|
if (type == Type::Float) {
|
||||||
array = OpConvertSToF(t_float, array);
|
array = OpConvertSToF(t_float, array);
|
||||||
|
@ -1767,7 +1767,7 @@ private:
|
||||||
operands.push_back(GetOffsetCoordinates(operation));
|
operands.push_back(GetOffsetCoordinates(operation));
|
||||||
}
|
}
|
||||||
|
|
||||||
if (meta.sampler.IsShadow()) {
|
if (meta.sampler.is_shadow) {
|
||||||
const Id dref = AsFloat(Visit(meta.depth_compare));
|
const Id dref = AsFloat(Visit(meta.depth_compare));
|
||||||
return {OpImageSampleDrefExplicitLod(t_float, sampler, coords, dref, mask, operands),
|
return {OpImageSampleDrefExplicitLod(t_float, sampler, coords, dref, mask, operands),
|
||||||
Type::Float};
|
Type::Float};
|
||||||
|
@ -1782,7 +1782,7 @@ private:
|
||||||
|
|
||||||
const Id coords = GetCoordinates(operation, Type::Float);
|
const Id coords = GetCoordinates(operation, Type::Float);
|
||||||
Id texture{};
|
Id texture{};
|
||||||
if (meta.sampler.IsShadow()) {
|
if (meta.sampler.is_shadow) {
|
||||||
texture = OpImageDrefGather(t_float4, GetTextureSampler(operation), coords,
|
texture = OpImageDrefGather(t_float4, GetTextureSampler(operation), coords,
|
||||||
AsFloat(Visit(meta.depth_compare)));
|
AsFloat(Visit(meta.depth_compare)));
|
||||||
} else {
|
} else {
|
||||||
|
@ -1809,8 +1809,8 @@ private:
|
||||||
}
|
}
|
||||||
|
|
||||||
const Id lod = AsUint(Visit(operation[0]));
|
const Id lod = AsUint(Visit(operation[0]));
|
||||||
const std::size_t coords_count = [&]() {
|
const std::size_t coords_count = [&meta] {
|
||||||
switch (const auto type = meta.sampler.GetType(); type) {
|
switch (const auto type = meta.sampler.type) {
|
||||||
case Tegra::Shader::TextureType::Texture1D:
|
case Tegra::Shader::TextureType::Texture1D:
|
||||||
return 1;
|
return 1;
|
||||||
case Tegra::Shader::TextureType::Texture2D:
|
case Tegra::Shader::TextureType::Texture2D:
|
||||||
|
@ -1819,7 +1819,7 @@ private:
|
||||||
case Tegra::Shader::TextureType::Texture3D:
|
case Tegra::Shader::TextureType::Texture3D:
|
||||||
return 3;
|
return 3;
|
||||||
default:
|
default:
|
||||||
UNREACHABLE_MSG("Invalid texture type={}", static_cast<u32>(type));
|
UNREACHABLE_MSG("Invalid texture type={}", static_cast<int>(type));
|
||||||
return 2;
|
return 2;
|
||||||
}
|
}
|
||||||
}();
|
}();
|
||||||
|
@ -1862,7 +1862,7 @@ private:
|
||||||
const Id image = GetTextureImage(operation);
|
const Id image = GetTextureImage(operation);
|
||||||
const Id coords = GetCoordinates(operation, Type::Int);
|
const Id coords = GetCoordinates(operation, Type::Int);
|
||||||
Id fetch;
|
Id fetch;
|
||||||
if (meta.lod && !meta.sampler.IsBuffer()) {
|
if (meta.lod && !meta.sampler.is_buffer) {
|
||||||
fetch = OpImageFetch(t_float4, image, coords, spv::ImageOperandsMask::Lod,
|
fetch = OpImageFetch(t_float4, image, coords, spv::ImageOperandsMask::Lod,
|
||||||
AsInt(Visit(meta.lod)));
|
AsInt(Visit(meta.lod)));
|
||||||
} else {
|
} else {
|
||||||
|
@ -2980,7 +2980,7 @@ ShaderEntries GenerateShaderEntries(const VideoCommon::Shader::ShaderIR& ir) {
|
||||||
entries.global_buffers.emplace_back(base.cbuf_index, base.cbuf_offset, usage.is_written);
|
entries.global_buffers.emplace_back(base.cbuf_index, base.cbuf_offset, usage.is_written);
|
||||||
}
|
}
|
||||||
for (const auto& sampler : ir.GetSamplers()) {
|
for (const auto& sampler : ir.GetSamplers()) {
|
||||||
if (sampler.IsBuffer()) {
|
if (sampler.is_buffer) {
|
||||||
entries.texel_buffers.emplace_back(sampler);
|
entries.texel_buffers.emplace_back(sampler);
|
||||||
} else {
|
} else {
|
||||||
entries.samplers.emplace_back(sampler);
|
entries.samplers.emplace_back(sampler);
|
||||||
|
|
|
@ -32,11 +32,11 @@ void DeduceTextureHandlerSize(VideoCore::GuestDriverProfile& gpu_driver,
|
||||||
u32 count{};
|
u32 count{};
|
||||||
std::vector<u32> bound_offsets;
|
std::vector<u32> bound_offsets;
|
||||||
for (const auto& sampler : used_samplers) {
|
for (const auto& sampler : used_samplers) {
|
||||||
if (sampler.IsBindless()) {
|
if (sampler.is_bindless) {
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
++count;
|
++count;
|
||||||
bound_offsets.emplace_back(sampler.GetOffset());
|
bound_offsets.emplace_back(sampler.offset);
|
||||||
}
|
}
|
||||||
if (count > 1) {
|
if (count > 1) {
|
||||||
gpu_driver.DeduceTextureHandlerSize(std::move(bound_offsets));
|
gpu_driver.DeduceTextureHandlerSize(std::move(bound_offsets));
|
||||||
|
@ -46,14 +46,14 @@ void DeduceTextureHandlerSize(VideoCore::GuestDriverProfile& gpu_driver,
|
||||||
std::optional<u32> TryDeduceSamplerSize(const Sampler& sampler_to_deduce,
|
std::optional<u32> TryDeduceSamplerSize(const Sampler& sampler_to_deduce,
|
||||||
VideoCore::GuestDriverProfile& gpu_driver,
|
VideoCore::GuestDriverProfile& gpu_driver,
|
||||||
const std::list<Sampler>& used_samplers) {
|
const std::list<Sampler>& used_samplers) {
|
||||||
const u32 base_offset = sampler_to_deduce.GetOffset();
|
const u32 base_offset = sampler_to_deduce.offset;
|
||||||
u32 max_offset{std::numeric_limits<u32>::max()};
|
u32 max_offset{std::numeric_limits<u32>::max()};
|
||||||
for (const auto& sampler : used_samplers) {
|
for (const auto& sampler : used_samplers) {
|
||||||
if (sampler.IsBindless()) {
|
if (sampler.is_bindless) {
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
if (sampler.GetOffset() > base_offset) {
|
if (sampler.offset > base_offset) {
|
||||||
max_offset = std::min(sampler.GetOffset(), max_offset);
|
max_offset = std::min(sampler.offset, max_offset);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
if (max_offset == std::numeric_limits<u32>::max()) {
|
if (max_offset == std::numeric_limits<u32>::max()) {
|
||||||
|
@ -353,14 +353,14 @@ void ShaderIR::PostDecode() {
|
||||||
return;
|
return;
|
||||||
}
|
}
|
||||||
for (auto& sampler : used_samplers) {
|
for (auto& sampler : used_samplers) {
|
||||||
if (!sampler.IsIndexed()) {
|
if (!sampler.is_indexed) {
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
if (const auto size = TryDeduceSamplerSize(sampler, gpu_driver, used_samplers)) {
|
if (const auto size = TryDeduceSamplerSize(sampler, gpu_driver, used_samplers)) {
|
||||||
sampler.SetSize(*size);
|
sampler.size = *size;
|
||||||
} else {
|
} else {
|
||||||
LOG_CRITICAL(HW_GPU, "Failed to deduce size of indexed sampler");
|
LOG_CRITICAL(HW_GPU, "Failed to deduce size of indexed sampler");
|
||||||
sampler.SetSize(1);
|
sampler.size = 1;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
|
@ -485,11 +485,10 @@ u32 ShaderIR::DecodeImage(NodeBlock& bb, u32 pc) {
|
||||||
Image& ShaderIR::GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType type) {
|
Image& ShaderIR::GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType type) {
|
||||||
const auto offset = static_cast<u32>(image.index.Value());
|
const auto offset = static_cast<u32>(image.index.Value());
|
||||||
|
|
||||||
const auto it =
|
const auto it = std::find_if(std::begin(used_images), std::end(used_images),
|
||||||
std::find_if(std::begin(used_images), std::end(used_images),
|
[offset](const Image& entry) { return entry.offset == offset; });
|
||||||
[offset](const Image& entry) { return entry.GetOffset() == offset; });
|
|
||||||
if (it != std::end(used_images)) {
|
if (it != std::end(used_images)) {
|
||||||
ASSERT(!it->IsBindless() && it->GetType() == it->GetType());
|
ASSERT(!it->is_bindless && it->type == type);
|
||||||
return *it;
|
return *it;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
@ -505,13 +504,12 @@ Image& ShaderIR::GetBindlessImage(Tegra::Shader::Register reg, Tegra::Shader::Im
|
||||||
const auto buffer = std::get<1>(result);
|
const auto buffer = std::get<1>(result);
|
||||||
const auto offset = std::get<2>(result);
|
const auto offset = std::get<2>(result);
|
||||||
|
|
||||||
const auto it =
|
const auto it = std::find_if(std::begin(used_images), std::end(used_images),
|
||||||
std::find_if(std::begin(used_images), std::end(used_images),
|
[buffer, offset](const Image& entry) {
|
||||||
[buffer = buffer, offset = offset](const Image& entry) {
|
return entry.buffer == buffer && entry.offset == offset;
|
||||||
return entry.GetBuffer() == buffer && entry.GetOffset() == offset;
|
|
||||||
});
|
});
|
||||||
if (it != std::end(used_images)) {
|
if (it != std::end(used_images)) {
|
||||||
ASSERT(it->IsBindless() && it->GetType() == it->GetType());
|
ASSERT(it->is_bindless && it->type == type);
|
||||||
return *it;
|
return *it;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
|
@ -139,15 +139,15 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
|
||||||
}
|
}
|
||||||
const Node component = Immediate(static_cast<u32>(instr.tld4s.component));
|
const Node component = Immediate(static_cast<u32>(instr.tld4s.component));
|
||||||
|
|
||||||
const SamplerInfo info{TextureType::Texture2D, false, is_depth_compare, false};
|
SamplerInfo info;
|
||||||
const Sampler& sampler = *GetSampler(instr.sampler, info);
|
info.is_shadow = is_depth_compare;
|
||||||
|
const std::optional<Sampler> sampler = GetSampler(instr.sampler, info);
|
||||||
|
|
||||||
Node4 values;
|
Node4 values;
|
||||||
for (u32 element = 0; element < values.size(); ++element) {
|
for (u32 element = 0; element < values.size(); ++element) {
|
||||||
auto coords_copy = coords;
|
MetaTexture meta{*sampler, {}, depth_compare, aoffi, {}, {},
|
||||||
MetaTexture meta{sampler, {}, depth_compare, aoffi, {}, {},
|
|
||||||
{}, {}, component, element, {}};
|
{}, {}, component, element, {}};
|
||||||
values[element] = Operation(OperationCode::TextureGather, meta, std::move(coords_copy));
|
values[element] = Operation(OperationCode::TextureGather, meta, coords);
|
||||||
}
|
}
|
||||||
|
|
||||||
if (instr.tld4s.fp16_flag) {
|
if (instr.tld4s.fp16_flag) {
|
||||||
|
@ -165,18 +165,20 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
|
||||||
"AOFFI is not implemented");
|
"AOFFI is not implemented");
|
||||||
|
|
||||||
const bool is_array = instr.txd.is_array != 0;
|
const bool is_array = instr.txd.is_array != 0;
|
||||||
u64 base_reg = instr.gpr8.Value();
|
|
||||||
const auto derivate_reg = instr.gpr20.Value();
|
const auto derivate_reg = instr.gpr20.Value();
|
||||||
const auto texture_type = instr.txd.texture_type.Value();
|
const auto texture_type = instr.txd.texture_type.Value();
|
||||||
const auto coord_count = GetCoordCount(texture_type);
|
const auto coord_count = GetCoordCount(texture_type);
|
||||||
Node index_var{};
|
u64 base_reg = instr.gpr8.Value();
|
||||||
const Sampler* sampler =
|
Node index_var;
|
||||||
is_bindless
|
SamplerInfo info;
|
||||||
? GetBindlessSampler(base_reg, index_var, {{texture_type, is_array, false, false}})
|
info.type = texture_type;
|
||||||
: GetSampler(instr.sampler, {{texture_type, is_array, false, false}});
|
info.is_array = is_array;
|
||||||
|
const std::optional<Sampler> sampler = is_bindless
|
||||||
|
? GetBindlessSampler(base_reg, info, index_var)
|
||||||
|
: GetSampler(instr.sampler, info);
|
||||||
Node4 values;
|
Node4 values;
|
||||||
if (sampler == nullptr) {
|
if (!sampler) {
|
||||||
std::generate(values.begin(), values.end(), [] { return Immediate(0); });
|
std::generate(values.begin(), values.end(), [this] { return Immediate(0); });
|
||||||
WriteTexInstructionFloat(bb, instr, values);
|
WriteTexInstructionFloat(bb, instr, values);
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
|
@ -214,14 +216,12 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
|
||||||
is_bindless = true;
|
is_bindless = true;
|
||||||
[[fallthrough]];
|
[[fallthrough]];
|
||||||
case OpCode::Id::TXQ: {
|
case OpCode::Id::TXQ: {
|
||||||
// TODO: The new commits on the texture refactor, change the way samplers work.
|
Node index_var;
|
||||||
// Sadly, not all texture instructions specify the type of texture their sampler
|
const std::optional<Sampler> sampler = is_bindless
|
||||||
// uses. This must be fixed at a later instance.
|
? GetBindlessSampler(instr.gpr8, {}, index_var)
|
||||||
Node index_var{};
|
: GetSampler(instr.sampler, {});
|
||||||
const Sampler* sampler =
|
|
||||||
is_bindless ? GetBindlessSampler(instr.gpr8, index_var) : GetSampler(instr.sampler);
|
|
||||||
|
|
||||||
if (sampler == nullptr) {
|
if (!sampler) {
|
||||||
u32 indexer = 0;
|
u32 indexer = 0;
|
||||||
for (u32 element = 0; element < 4; ++element) {
|
for (u32 element = 0; element < 4; ++element) {
|
||||||
if (!instr.txq.IsComponentEnabled(element)) {
|
if (!instr.txq.IsComponentEnabled(element)) {
|
||||||
|
@ -267,12 +267,17 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
|
||||||
UNIMPLEMENTED_IF_MSG(instr.tmml.UsesMiscMode(Tegra::Shader::TextureMiscMode::NDV),
|
UNIMPLEMENTED_IF_MSG(instr.tmml.UsesMiscMode(Tegra::Shader::TextureMiscMode::NDV),
|
||||||
"NDV is not implemented");
|
"NDV is not implemented");
|
||||||
|
|
||||||
auto texture_type = instr.tmml.texture_type.Value();
|
const auto texture_type = instr.tmml.texture_type.Value();
|
||||||
Node index_var{};
|
const bool is_array = instr.tmml.array != 0;
|
||||||
const Sampler* sampler =
|
SamplerInfo info;
|
||||||
is_bindless ? GetBindlessSampler(instr.gpr20, index_var) : GetSampler(instr.sampler);
|
info.type = texture_type;
|
||||||
|
info.is_array = is_array;
|
||||||
|
Node index_var;
|
||||||
|
const std::optional<Sampler> sampler =
|
||||||
|
is_bindless ? GetBindlessSampler(instr.gpr20, info, index_var)
|
||||||
|
: GetSampler(instr.sampler, info);
|
||||||
|
|
||||||
if (sampler == nullptr) {
|
if (!sampler) {
|
||||||
u32 indexer = 0;
|
u32 indexer = 0;
|
||||||
for (u32 element = 0; element < 2; ++element) {
|
for (u32 element = 0; element < 2; ++element) {
|
||||||
if (!instr.tmml.IsComponentEnabled(element)) {
|
if (!instr.tmml.IsComponentEnabled(element)) {
|
||||||
|
@ -299,12 +304,11 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
|
||||||
coords.push_back(GetRegister(instr.gpr8.Value() + 1));
|
coords.push_back(GetRegister(instr.gpr8.Value() + 1));
|
||||||
break;
|
break;
|
||||||
default:
|
default:
|
||||||
UNIMPLEMENTED_MSG("Unhandled texture type {}", static_cast<u32>(texture_type));
|
UNIMPLEMENTED_MSG("Unhandled texture type {}", static_cast<int>(texture_type));
|
||||||
|
|
||||||
// Fallback to interpreting as a 2D texture for now
|
// Fallback to interpreting as a 2D texture for now
|
||||||
coords.push_back(GetRegister(instr.gpr8.Value() + 0));
|
coords.push_back(GetRegister(instr.gpr8.Value() + 0));
|
||||||
coords.push_back(GetRegister(instr.gpr8.Value() + 1));
|
coords.push_back(GetRegister(instr.gpr8.Value() + 1));
|
||||||
texture_type = TextureType::Texture2D;
|
|
||||||
}
|
}
|
||||||
u32 indexer = 0;
|
u32 indexer = 0;
|
||||||
for (u32 element = 0; element < 2; ++element) {
|
for (u32 element = 0; element < 2; ++element) {
|
||||||
|
@ -353,98 +357,103 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
|
||||||
return pc;
|
return pc;
|
||||||
}
|
}
|
||||||
|
|
||||||
ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo(std::optional<SamplerInfo> sampler_info, u32 offset,
|
ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo(SamplerInfo info, u32 offset,
|
||||||
std::optional<u32> buffer) {
|
std::optional<u32> buffer) {
|
||||||
if (sampler_info) {
|
if (info.IsComplete()) {
|
||||||
return *sampler_info;
|
return info;
|
||||||
}
|
}
|
||||||
const auto sampler = buffer ? registry.ObtainBindlessSampler(*buffer, offset)
|
const auto sampler = buffer ? registry.ObtainBindlessSampler(*buffer, offset)
|
||||||
: registry.ObtainBoundSampler(offset);
|
: registry.ObtainBoundSampler(offset);
|
||||||
if (!sampler) {
|
if (!sampler) {
|
||||||
LOG_WARNING(HW_GPU, "Unknown sampler info");
|
LOG_WARNING(HW_GPU, "Unknown sampler info");
|
||||||
return SamplerInfo{TextureType::Texture2D, false, false, false};
|
info.type = info.type.value_or(Tegra::Shader::TextureType::Texture2D);
|
||||||
|
info.is_array = info.is_array.value_or(false);
|
||||||
|
info.is_shadow = info.is_shadow.value_or(false);
|
||||||
|
info.is_buffer = info.is_buffer.value_or(false);
|
||||||
|
return info;
|
||||||
}
|
}
|
||||||
return SamplerInfo{sampler->texture_type, sampler->is_array != 0, sampler->is_shadow != 0,
|
info.type = info.type.value_or(sampler->texture_type);
|
||||||
sampler->is_buffer != 0};
|
info.is_array = info.is_array.value_or(sampler->is_array != 0);
|
||||||
|
info.is_shadow = info.is_shadow.value_or(sampler->is_shadow != 0);
|
||||||
|
info.is_buffer = info.is_buffer.value_or(sampler->is_buffer != 0);
|
||||||
|
return info;
|
||||||
}
|
}
|
||||||
|
|
||||||
const Sampler* ShaderIR::GetSampler(const Tegra::Shader::Sampler& sampler,
|
std::optional<Sampler> ShaderIR::GetSampler(Tegra::Shader::Sampler sampler,
|
||||||
std::optional<SamplerInfo> sampler_info) {
|
SamplerInfo sampler_info) {
|
||||||
const auto offset = static_cast<u32>(sampler.index.Value());
|
const auto offset = static_cast<u32>(sampler.index.Value());
|
||||||
const auto info = GetSamplerInfo(sampler_info, offset);
|
const auto info = GetSamplerInfo(sampler_info, offset);
|
||||||
|
|
||||||
// If this sampler has already been used, return the existing mapping.
|
// If this sampler has already been used, return the existing mapping.
|
||||||
const auto it =
|
const auto it = std::find_if(used_samplers.begin(), used_samplers.end(),
|
||||||
std::find_if(used_samplers.begin(), used_samplers.end(),
|
[offset](const Sampler& entry) { return entry.offset == offset; });
|
||||||
[offset](const Sampler& entry) { return entry.GetOffset() == offset; });
|
|
||||||
if (it != used_samplers.end()) {
|
if (it != used_samplers.end()) {
|
||||||
ASSERT(!it->IsBindless() && it->GetType() == info.type && it->IsArray() == info.is_array &&
|
ASSERT(!it->is_bindless && it->type == info.type && it->is_array == info.is_array &&
|
||||||
it->IsShadow() == info.is_shadow && it->IsBuffer() == info.is_buffer);
|
it->is_shadow == info.is_shadow && it->is_buffer == info.is_buffer);
|
||||||
return &*it;
|
return *it;
|
||||||
}
|
}
|
||||||
|
|
||||||
// Otherwise create a new mapping for this sampler
|
// Otherwise create a new mapping for this sampler
|
||||||
const auto next_index = static_cast<u32>(used_samplers.size());
|
const auto next_index = static_cast<u32>(used_samplers.size());
|
||||||
return &used_samplers.emplace_back(next_index, offset, info.type, info.is_array, info.is_shadow,
|
return used_samplers.emplace_back(next_index, offset, *info.type, *info.is_array,
|
||||||
info.is_buffer, false);
|
*info.is_shadow, *info.is_buffer, false);
|
||||||
}
|
}
|
||||||
|
|
||||||
const Sampler* ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg, Node& index_var,
|
std::optional<Sampler> ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg, SamplerInfo info,
|
||||||
std::optional<SamplerInfo> sampler_info) {
|
Node& index_var) {
|
||||||
const Node sampler_register = GetRegister(reg);
|
const Node sampler_register = GetRegister(reg);
|
||||||
const auto [base_node, tracked_sampler_info] =
|
const auto [base_node, tracked_sampler_info] =
|
||||||
TrackBindlessSampler(sampler_register, global_code, static_cast<s64>(global_code.size()));
|
TrackBindlessSampler(sampler_register, global_code, static_cast<s64>(global_code.size()));
|
||||||
ASSERT(base_node != nullptr);
|
ASSERT(base_node != nullptr);
|
||||||
if (base_node == nullptr) {
|
if (base_node == nullptr) {
|
||||||
return nullptr;
|
return std::nullopt;
|
||||||
}
|
}
|
||||||
|
|
||||||
if (const auto bindless_sampler_info =
|
if (const auto bindless_sampler_info =
|
||||||
std::get_if<BindlessSamplerNode>(&*tracked_sampler_info)) {
|
std::get_if<BindlessSamplerNode>(&*tracked_sampler_info)) {
|
||||||
const u32 buffer = bindless_sampler_info->GetIndex();
|
const u32 buffer = bindless_sampler_info->GetIndex();
|
||||||
const u32 offset = bindless_sampler_info->GetOffset();
|
const u32 offset = bindless_sampler_info->GetOffset();
|
||||||
const auto info = GetSamplerInfo(sampler_info, offset, buffer);
|
info = GetSamplerInfo(info, offset, buffer);
|
||||||
|
|
||||||
// If this sampler has already been used, return the existing mapping.
|
// If this sampler has already been used, return the existing mapping.
|
||||||
const auto it =
|
const auto it = std::find_if(used_samplers.begin(), used_samplers.end(),
|
||||||
std::find_if(used_samplers.begin(), used_samplers.end(),
|
|
||||||
[buffer = buffer, offset = offset](const Sampler& entry) {
|
[buffer = buffer, offset = offset](const Sampler& entry) {
|
||||||
return entry.GetBuffer() == buffer && entry.GetOffset() == offset;
|
return entry.buffer == buffer && entry.offset == offset;
|
||||||
});
|
});
|
||||||
if (it != used_samplers.end()) {
|
if (it != used_samplers.end()) {
|
||||||
ASSERT(it->IsBindless() && it->GetType() == info.type &&
|
ASSERT(it->is_bindless && it->type == info.type && it->is_array == info.is_array &&
|
||||||
it->IsArray() == info.is_array && it->IsShadow() == info.is_shadow);
|
it->is_shadow == info.is_shadow);
|
||||||
return &*it;
|
return *it;
|
||||||
}
|
}
|
||||||
|
|
||||||
// Otherwise create a new mapping for this sampler
|
// Otherwise create a new mapping for this sampler
|
||||||
const auto next_index = static_cast<u32>(used_samplers.size());
|
const auto next_index = static_cast<u32>(used_samplers.size());
|
||||||
return &used_samplers.emplace_back(next_index, offset, buffer, info.type, info.is_array,
|
return used_samplers.emplace_back(next_index, offset, buffer, *info.type, *info.is_array,
|
||||||
info.is_shadow, info.is_buffer, false);
|
*info.is_shadow, *info.is_buffer, false);
|
||||||
} else if (const auto array_sampler_info =
|
}
|
||||||
std::get_if<ArraySamplerNode>(&*tracked_sampler_info)) {
|
if (const auto array_sampler_info = std::get_if<ArraySamplerNode>(&*tracked_sampler_info)) {
|
||||||
const u32 base_offset = array_sampler_info->GetBaseOffset() / 4;
|
const u32 base_offset = array_sampler_info->GetBaseOffset() / 4;
|
||||||
index_var = GetCustomVariable(array_sampler_info->GetIndexVar());
|
index_var = GetCustomVariable(array_sampler_info->GetIndexVar());
|
||||||
const auto info = GetSamplerInfo(sampler_info, base_offset);
|
info = GetSamplerInfo(info, base_offset);
|
||||||
|
|
||||||
// If this sampler has already been used, return the existing mapping.
|
// If this sampler has already been used, return the existing mapping.
|
||||||
const auto it = std::find_if(
|
const auto it = std::find_if(
|
||||||
used_samplers.begin(), used_samplers.end(),
|
used_samplers.begin(), used_samplers.end(),
|
||||||
[base_offset](const Sampler& entry) { return entry.GetOffset() == base_offset; });
|
[base_offset](const Sampler& entry) { return entry.offset == base_offset; });
|
||||||
if (it != used_samplers.end()) {
|
if (it != used_samplers.end()) {
|
||||||
ASSERT(!it->IsBindless() && it->GetType() == info.type &&
|
ASSERT(!it->is_bindless && it->type == info.type && it->is_array == info.is_array &&
|
||||||
it->IsArray() == info.is_array && it->IsShadow() == info.is_shadow &&
|
it->is_shadow == info.is_shadow && it->is_buffer == info.is_buffer &&
|
||||||
it->IsBuffer() == info.is_buffer && it->IsIndexed());
|
it->is_indexed);
|
||||||
return &*it;
|
return *it;
|
||||||
}
|
}
|
||||||
|
|
||||||
uses_indexed_samplers = true;
|
uses_indexed_samplers = true;
|
||||||
// Otherwise create a new mapping for this sampler
|
// Otherwise create a new mapping for this sampler
|
||||||
const auto next_index = static_cast<u32>(used_samplers.size());
|
const auto next_index = static_cast<u32>(used_samplers.size());
|
||||||
return &used_samplers.emplace_back(next_index, base_offset, info.type, info.is_array,
|
return used_samplers.emplace_back(next_index, base_offset, *info.type, *info.is_array,
|
||||||
info.is_shadow, info.is_buffer, true);
|
*info.is_shadow, *info.is_buffer, true);
|
||||||
}
|
}
|
||||||
return nullptr;
|
return std::nullopt;
|
||||||
}
|
}
|
||||||
|
|
||||||
void ShaderIR::WriteTexInstructionFloat(NodeBlock& bb, Instruction instr, const Node4& components) {
|
void ShaderIR::WriteTexInstructionFloat(NodeBlock& bb, Instruction instr, const Node4& components) {
|
||||||
|
@ -529,9 +538,15 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
|
||||||
ASSERT_MSG(texture_type != TextureType::Texture3D || !is_array || !is_shadow,
|
ASSERT_MSG(texture_type != TextureType::Texture3D || !is_array || !is_shadow,
|
||||||
"Illegal texture type");
|
"Illegal texture type");
|
||||||
|
|
||||||
const SamplerInfo info{texture_type, is_array, is_shadow, false};
|
SamplerInfo info;
|
||||||
|
info.type = texture_type;
|
||||||
|
info.is_array = is_array;
|
||||||
|
info.is_shadow = is_shadow;
|
||||||
|
info.is_buffer = false;
|
||||||
|
|
||||||
Node index_var;
|
Node index_var;
|
||||||
const Sampler* sampler = is_bindless ? GetBindlessSampler(*bindless_reg, index_var, info)
|
const std::optional<Sampler> sampler = is_bindless
|
||||||
|
? GetBindlessSampler(*bindless_reg, info, index_var)
|
||||||
: GetSampler(instr.sampler, info);
|
: GetSampler(instr.sampler, info);
|
||||||
if (!sampler) {
|
if (!sampler) {
|
||||||
return {Immediate(0), Immediate(0), Immediate(0), Immediate(0)};
|
return {Immediate(0), Immediate(0), Immediate(0), Immediate(0)};
|
||||||
|
@ -683,12 +698,17 @@ Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool de
|
||||||
|
|
||||||
u64 parameter_register = instr.gpr20.Value();
|
u64 parameter_register = instr.gpr20.Value();
|
||||||
|
|
||||||
const SamplerInfo info{texture_type, is_array, depth_compare, false};
|
SamplerInfo info;
|
||||||
Node index_var{};
|
info.type = texture_type;
|
||||||
const Sampler* sampler = is_bindless ? GetBindlessSampler(parameter_register++, index_var, info)
|
info.is_array = is_array;
|
||||||
|
info.is_shadow = depth_compare;
|
||||||
|
|
||||||
|
Node index_var;
|
||||||
|
const std::optional<Sampler> sampler =
|
||||||
|
is_bindless ? GetBindlessSampler(parameter_register++, info, index_var)
|
||||||
: GetSampler(instr.sampler, info);
|
: GetSampler(instr.sampler, info);
|
||||||
Node4 values;
|
Node4 values;
|
||||||
if (sampler == nullptr) {
|
if (!sampler) {
|
||||||
for (u32 element = 0; element < values.size(); ++element) {
|
for (u32 element = 0; element < values.size(); ++element) {
|
||||||
values[element] = Immediate(0);
|
values[element] = Immediate(0);
|
||||||
}
|
}
|
||||||
|
@ -743,12 +763,12 @@ Node4 ShaderIR::GetTldCode(Tegra::Shader::Instruction instr) {
|
||||||
// const Node aoffi_register{is_aoffi ? GetRegister(gpr20_cursor++) : nullptr};
|
// const Node aoffi_register{is_aoffi ? GetRegister(gpr20_cursor++) : nullptr};
|
||||||
// const Node multisample{is_multisample ? GetRegister(gpr20_cursor++) : nullptr};
|
// const Node multisample{is_multisample ? GetRegister(gpr20_cursor++) : nullptr};
|
||||||
|
|
||||||
const auto& sampler = *GetSampler(instr.sampler);
|
const std::optional<Sampler> sampler = GetSampler(instr.sampler, {});
|
||||||
|
|
||||||
Node4 values;
|
Node4 values;
|
||||||
for (u32 element = 0; element < values.size(); ++element) {
|
for (u32 element = 0; element < values.size(); ++element) {
|
||||||
auto coords_copy = coords;
|
auto coords_copy = coords;
|
||||||
MetaTexture meta{sampler, array_register, {}, {}, {}, {}, {}, lod, {}, element, {}};
|
MetaTexture meta{*sampler, array_register, {}, {}, {}, {}, {}, lod, {}, element, {}};
|
||||||
values[element] = Operation(OperationCode::TexelFetch, meta, std::move(coords_copy));
|
values[element] = Operation(OperationCode::TexelFetch, meta, std::move(coords_copy));
|
||||||
}
|
}
|
||||||
|
|
||||||
|
@ -756,7 +776,11 @@ Node4 ShaderIR::GetTldCode(Tegra::Shader::Instruction instr) {
|
||||||
}
|
}
|
||||||
|
|
||||||
Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is_array) {
|
Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is_array) {
|
||||||
const Sampler& sampler = *GetSampler(instr.sampler);
|
SamplerInfo info;
|
||||||
|
info.type = texture_type;
|
||||||
|
info.is_array = is_array;
|
||||||
|
info.is_shadow = false;
|
||||||
|
const std::optional<Sampler> sampler = GetSampler(instr.sampler, info);
|
||||||
|
|
||||||
const std::size_t type_coord_count = GetCoordCount(texture_type);
|
const std::size_t type_coord_count = GetCoordCount(texture_type);
|
||||||
const bool lod_enabled = instr.tlds.GetTextureProcessMode() == TextureProcessMode::LL;
|
const bool lod_enabled = instr.tlds.GetTextureProcessMode() == TextureProcessMode::LL;
|
||||||
|
@ -784,7 +808,7 @@ Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is
|
||||||
Node4 values;
|
Node4 values;
|
||||||
for (u32 element = 0; element < values.size(); ++element) {
|
for (u32 element = 0; element < values.size(); ++element) {
|
||||||
auto coords_copy = coords;
|
auto coords_copy = coords;
|
||||||
MetaTexture meta{sampler, array, {}, {}, {}, {}, {}, lod, {}, element, {}};
|
MetaTexture meta{*sampler, array, {}, {}, {}, {}, {}, lod, {}, element, {}};
|
||||||
values[element] = Operation(OperationCode::TexelFetch, meta, std::move(coords_copy));
|
values[element] = Operation(OperationCode::TexelFetch, meta, std::move(coords_copy));
|
||||||
}
|
}
|
||||||
return values;
|
return values;
|
||||||
|
|
|
@ -267,76 +267,30 @@ class ArraySamplerNode;
|
||||||
using TrackSamplerData = std::variant<BindlessSamplerNode, ArraySamplerNode>;
|
using TrackSamplerData = std::variant<BindlessSamplerNode, ArraySamplerNode>;
|
||||||
using TrackSampler = std::shared_ptr<TrackSamplerData>;
|
using TrackSampler = std::shared_ptr<TrackSamplerData>;
|
||||||
|
|
||||||
class Sampler {
|
struct Sampler {
|
||||||
public:
|
/// Bound samplers constructor
|
||||||
/// This constructor is for bound samplers
|
|
||||||
constexpr explicit Sampler(u32 index, u32 offset, Tegra::Shader::TextureType type,
|
constexpr explicit Sampler(u32 index, u32 offset, Tegra::Shader::TextureType type,
|
||||||
bool is_array, bool is_shadow, bool is_buffer, bool is_indexed)
|
bool is_array, bool is_shadow, bool is_buffer, bool is_indexed)
|
||||||
: index{index}, offset{offset}, type{type}, is_array{is_array}, is_shadow{is_shadow},
|
: index{index}, offset{offset}, type{type}, is_array{is_array}, is_shadow{is_shadow},
|
||||||
is_buffer{is_buffer}, is_indexed{is_indexed} {}
|
is_buffer{is_buffer}, is_indexed{is_indexed} {}
|
||||||
|
|
||||||
/// This constructor is for bindless samplers
|
/// Bindless samplers constructor
|
||||||
constexpr explicit Sampler(u32 index, u32 offset, u32 buffer, Tegra::Shader::TextureType type,
|
constexpr explicit Sampler(u32 index, u32 offset, u32 buffer, Tegra::Shader::TextureType type,
|
||||||
bool is_array, bool is_shadow, bool is_buffer, bool is_indexed)
|
bool is_array, bool is_shadow, bool is_buffer, bool is_indexed)
|
||||||
: index{index}, offset{offset}, buffer{buffer}, type{type}, is_array{is_array},
|
: index{index}, offset{offset}, buffer{buffer}, type{type}, is_array{is_array},
|
||||||
is_shadow{is_shadow}, is_buffer{is_buffer}, is_bindless{true}, is_indexed{is_indexed} {}
|
is_shadow{is_shadow}, is_buffer{is_buffer}, is_bindless{true}, is_indexed{is_indexed} {}
|
||||||
|
|
||||||
constexpr u32 GetIndex() const {
|
u32 index = 0; ///< Emulated index given for the this sampler.
|
||||||
return index;
|
u32 offset = 0; ///< Offset in the const buffer from where the sampler is being read.
|
||||||
}
|
u32 buffer = 0; ///< Buffer where the bindless sampler is being read (unused on bound samplers).
|
||||||
|
u32 size = 1; ///< Size of the sampler.
|
||||||
constexpr u32 GetOffset() const {
|
|
||||||
return offset;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr u32 GetBuffer() const {
|
|
||||||
return buffer;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr Tegra::Shader::TextureType GetType() const {
|
|
||||||
return type;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr bool IsArray() const {
|
|
||||||
return is_array;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr bool IsShadow() const {
|
|
||||||
return is_shadow;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr bool IsBuffer() const {
|
|
||||||
return is_buffer;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr bool IsBindless() const {
|
|
||||||
return is_bindless;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr bool IsIndexed() const {
|
|
||||||
return is_indexed;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr u32 Size() const {
|
|
||||||
return size;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr void SetSize(u32 new_size) {
|
|
||||||
size = new_size;
|
|
||||||
}
|
|
||||||
|
|
||||||
private:
|
|
||||||
u32 index{}; ///< Emulated index given for the this sampler.
|
|
||||||
u32 offset{}; ///< Offset in the const buffer from where the sampler is being read.
|
|
||||||
u32 buffer{}; ///< Buffer where the bindless sampler is being read (unused on bound samplers).
|
|
||||||
u32 size{1}; ///< Size of the sampler.
|
|
||||||
|
|
||||||
Tegra::Shader::TextureType type{}; ///< The type used to sample this texture (Texture2D, etc)
|
Tegra::Shader::TextureType type{}; ///< The type used to sample this texture (Texture2D, etc)
|
||||||
bool is_array{}; ///< Whether the texture is being sampled as an array texture or not.
|
bool is_array = false; ///< Whether the texture is being sampled as an array texture or not.
|
||||||
bool is_shadow{}; ///< Whether the texture is being sampled as a depth texture or not.
|
bool is_shadow = false; ///< Whether the texture is being sampled as a depth texture or not.
|
||||||
bool is_buffer{}; ///< Whether the texture is a texture buffer without sampler.
|
bool is_buffer = false; ///< Whether the texture is a texture buffer without sampler.
|
||||||
bool is_bindless{}; ///< Whether this sampler belongs to a bindless texture or not.
|
bool is_bindless = false; ///< Whether this sampler belongs to a bindless texture or not.
|
||||||
bool is_indexed{}; ///< Whether this sampler is an indexed array of textures.
|
bool is_indexed = false; ///< Whether this sampler is an indexed array of textures.
|
||||||
};
|
};
|
||||||
|
|
||||||
/// Represents a tracked bindless sampler into a direct const buffer
|
/// Represents a tracked bindless sampler into a direct const buffer
|
||||||
|
@ -381,13 +335,13 @@ private:
|
||||||
u32 offset;
|
u32 offset;
|
||||||
};
|
};
|
||||||
|
|
||||||
class Image final {
|
struct Image {
|
||||||
public:
|
public:
|
||||||
/// This constructor is for bound images
|
/// Bound images constructor
|
||||||
constexpr explicit Image(u32 index, u32 offset, Tegra::Shader::ImageType type)
|
constexpr explicit Image(u32 index, u32 offset, Tegra::Shader::ImageType type)
|
||||||
: index{index}, offset{offset}, type{type} {}
|
: index{index}, offset{offset}, type{type} {}
|
||||||
|
|
||||||
/// This constructor is for bindless samplers
|
/// Bindless samplers constructor
|
||||||
constexpr explicit Image(u32 index, u32 offset, u32 buffer, Tegra::Shader::ImageType type)
|
constexpr explicit Image(u32 index, u32 offset, u32 buffer, Tegra::Shader::ImageType type)
|
||||||
: index{index}, offset{offset}, buffer{buffer}, type{type}, is_bindless{true} {}
|
: index{index}, offset{offset}, buffer{buffer}, type{type}, is_bindless{true} {}
|
||||||
|
|
||||||
|
@ -405,53 +359,20 @@ public:
|
||||||
is_atomic = true;
|
is_atomic = true;
|
||||||
}
|
}
|
||||||
|
|
||||||
constexpr u32 GetIndex() const {
|
u32 index = 0;
|
||||||
return index;
|
u32 offset = 0;
|
||||||
}
|
u32 buffer = 0;
|
||||||
|
|
||||||
constexpr u32 GetOffset() const {
|
|
||||||
return offset;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr u32 GetBuffer() const {
|
|
||||||
return buffer;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr Tegra::Shader::ImageType GetType() const {
|
|
||||||
return type;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr bool IsBindless() const {
|
|
||||||
return is_bindless;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr bool IsWritten() const {
|
|
||||||
return is_written;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr bool IsRead() const {
|
|
||||||
return is_read;
|
|
||||||
}
|
|
||||||
|
|
||||||
constexpr bool IsAtomic() const {
|
|
||||||
return is_atomic;
|
|
||||||
}
|
|
||||||
|
|
||||||
private:
|
|
||||||
u32 index{};
|
|
||||||
u32 offset{};
|
|
||||||
u32 buffer{};
|
|
||||||
|
|
||||||
Tegra::Shader::ImageType type{};
|
Tegra::Shader::ImageType type{};
|
||||||
bool is_bindless{};
|
bool is_bindless = false;
|
||||||
bool is_written{};
|
bool is_written = false;
|
||||||
bool is_read{};
|
bool is_read = false;
|
||||||
bool is_atomic{};
|
bool is_atomic = false;
|
||||||
};
|
};
|
||||||
|
|
||||||
struct GlobalMemoryBase {
|
struct GlobalMemoryBase {
|
||||||
u32 cbuf_index{};
|
u32 cbuf_index = 0;
|
||||||
u32 cbuf_offset{};
|
u32 cbuf_offset = 0;
|
||||||
|
|
||||||
bool operator<(const GlobalMemoryBase& rhs) const {
|
bool operator<(const GlobalMemoryBase& rhs) const {
|
||||||
return std::tie(cbuf_index, cbuf_offset) < std::tie(rhs.cbuf_index, rhs.cbuf_offset);
|
return std::tie(cbuf_index, cbuf_offset) < std::tie(rhs.cbuf_index, rhs.cbuf_offset);
|
||||||
|
@ -465,7 +386,7 @@ struct MetaArithmetic {
|
||||||
|
|
||||||
/// Parameters describing a texture sampler
|
/// Parameters describing a texture sampler
|
||||||
struct MetaTexture {
|
struct MetaTexture {
|
||||||
const Sampler& sampler;
|
Sampler sampler;
|
||||||
Node array;
|
Node array;
|
||||||
Node depth_compare;
|
Node depth_compare;
|
||||||
std::vector<Node> aoffi;
|
std::vector<Node> aoffi;
|
||||||
|
|
|
@ -28,12 +28,11 @@ struct ShaderBlock;
|
||||||
|
|
||||||
constexpr u32 MAX_PROGRAM_LENGTH = 0x1000;
|
constexpr u32 MAX_PROGRAM_LENGTH = 0x1000;
|
||||||
|
|
||||||
class ConstBuffer {
|
struct ConstBuffer {
|
||||||
public:
|
constexpr explicit ConstBuffer(u32 max_offset, bool is_indirect)
|
||||||
explicit ConstBuffer(u32 max_offset, bool is_indirect)
|
|
||||||
: max_offset{max_offset}, is_indirect{is_indirect} {}
|
: max_offset{max_offset}, is_indirect{is_indirect} {}
|
||||||
|
|
||||||
ConstBuffer() = default;
|
constexpr ConstBuffer() = default;
|
||||||
|
|
||||||
void MarkAsUsed(u64 offset) {
|
void MarkAsUsed(u64 offset) {
|
||||||
max_offset = std::max(max_offset, static_cast<u32>(offset));
|
max_offset = std::max(max_offset, static_cast<u32>(offset));
|
||||||
|
@ -56,8 +55,8 @@ public:
|
||||||
}
|
}
|
||||||
|
|
||||||
private:
|
private:
|
||||||
u32 max_offset{};
|
u32 max_offset = 0;
|
||||||
bool is_indirect{};
|
bool is_indirect = false;
|
||||||
};
|
};
|
||||||
|
|
||||||
struct GlobalMemoryUsage {
|
struct GlobalMemoryUsage {
|
||||||
|
@ -191,10 +190,14 @@ private:
|
||||||
friend class ASTDecoder;
|
friend class ASTDecoder;
|
||||||
|
|
||||||
struct SamplerInfo {
|
struct SamplerInfo {
|
||||||
Tegra::Shader::TextureType type;
|
std::optional<Tegra::Shader::TextureType> type;
|
||||||
bool is_array;
|
std::optional<bool> is_array;
|
||||||
bool is_shadow;
|
std::optional<bool> is_shadow;
|
||||||
bool is_buffer;
|
std::optional<bool> is_buffer;
|
||||||
|
|
||||||
|
constexpr bool IsComplete() const noexcept {
|
||||||
|
return type && is_array && is_shadow && is_buffer;
|
||||||
|
}
|
||||||
};
|
};
|
||||||
|
|
||||||
void Decode();
|
void Decode();
|
||||||
|
@ -327,16 +330,15 @@ private:
|
||||||
OperationCode GetPredicateCombiner(Tegra::Shader::PredOperation operation);
|
OperationCode GetPredicateCombiner(Tegra::Shader::PredOperation operation);
|
||||||
|
|
||||||
/// Queries the missing sampler info from the execution context.
|
/// Queries the missing sampler info from the execution context.
|
||||||
SamplerInfo GetSamplerInfo(std::optional<SamplerInfo> sampler_info, u32 offset,
|
SamplerInfo GetSamplerInfo(SamplerInfo info, u32 offset,
|
||||||
std::optional<u32> buffer = std::nullopt);
|
std::optional<u32> buffer = std::nullopt);
|
||||||
|
|
||||||
/// Accesses a texture sampler
|
/// Accesses a texture sampler.
|
||||||
const Sampler* GetSampler(const Tegra::Shader::Sampler& sampler,
|
std::optional<Sampler> GetSampler(Tegra::Shader::Sampler sampler, SamplerInfo info);
|
||||||
std::optional<SamplerInfo> sampler_info = std::nullopt);
|
|
||||||
|
|
||||||
/// Accesses a texture sampler for a bindless texture.
|
/// Accesses a texture sampler for a bindless texture.
|
||||||
const Sampler* GetBindlessSampler(Tegra::Shader::Register reg, Node& index_var,
|
std::optional<Sampler> GetBindlessSampler(Tegra::Shader::Register reg, SamplerInfo info,
|
||||||
std::optional<SamplerInfo> sampler_info = std::nullopt);
|
Node& index_var);
|
||||||
|
|
||||||
/// Accesses an image.
|
/// Accesses an image.
|
||||||
Image& GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType type);
|
Image& GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType type);
|
||||||
|
|
|
@ -81,7 +81,7 @@ SurfaceParams SurfaceParams::CreateForTexture(const FormatLookupTable& lookup_ta
|
||||||
params.pixel_format = lookup_table.GetPixelFormat(
|
params.pixel_format = lookup_table.GetPixelFormat(
|
||||||
tic.format, params.srgb_conversion, tic.r_type, tic.g_type, tic.b_type, tic.a_type);
|
tic.format, params.srgb_conversion, tic.r_type, tic.g_type, tic.b_type, tic.a_type);
|
||||||
params.type = GetFormatType(params.pixel_format);
|
params.type = GetFormatType(params.pixel_format);
|
||||||
if (entry.IsShadow() && params.type == SurfaceType::ColorTexture) {
|
if (entry.is_shadow && params.type == SurfaceType::ColorTexture) {
|
||||||
switch (params.pixel_format) {
|
switch (params.pixel_format) {
|
||||||
case PixelFormat::R16U:
|
case PixelFormat::R16U:
|
||||||
case PixelFormat::R16F:
|
case PixelFormat::R16F:
|
||||||
|
@ -108,7 +108,7 @@ SurfaceParams SurfaceParams::CreateForTexture(const FormatLookupTable& lookup_ta
|
||||||
params.emulated_levels = 1;
|
params.emulated_levels = 1;
|
||||||
params.is_layered = false;
|
params.is_layered = false;
|
||||||
} else {
|
} else {
|
||||||
params.target = TextureTypeToSurfaceTarget(entry.GetType(), entry.IsArray());
|
params.target = TextureTypeToSurfaceTarget(entry.type, entry.is_array);
|
||||||
params.width = tic.Width();
|
params.width = tic.Width();
|
||||||
params.height = tic.Height();
|
params.height = tic.Height();
|
||||||
params.depth = tic.Depth();
|
params.depth = tic.Depth();
|
||||||
|
@ -138,7 +138,7 @@ SurfaceParams SurfaceParams::CreateForImage(const FormatLookupTable& lookup_tabl
|
||||||
tic.format, params.srgb_conversion, tic.r_type, tic.g_type, tic.b_type, tic.a_type);
|
tic.format, params.srgb_conversion, tic.r_type, tic.g_type, tic.b_type, tic.a_type);
|
||||||
params.type = GetFormatType(params.pixel_format);
|
params.type = GetFormatType(params.pixel_format);
|
||||||
params.type = GetFormatType(params.pixel_format);
|
params.type = GetFormatType(params.pixel_format);
|
||||||
params.target = ImageTypeToSurfaceTarget(entry.GetType());
|
params.target = ImageTypeToSurfaceTarget(entry.type);
|
||||||
// TODO: on 1DBuffer we should use the tic info.
|
// TODO: on 1DBuffer we should use the tic info.
|
||||||
if (tic.IsBuffer()) {
|
if (tic.IsBuffer()) {
|
||||||
params.target = SurfaceTarget::TextureBuffer;
|
params.target = SurfaceTarget::TextureBuffer;
|
||||||
|
@ -248,12 +248,12 @@ SurfaceParams SurfaceParams::CreateForFermiCopySurface(
|
||||||
|
|
||||||
VideoCore::Surface::SurfaceTarget SurfaceParams::ExpectedTarget(
|
VideoCore::Surface::SurfaceTarget SurfaceParams::ExpectedTarget(
|
||||||
const VideoCommon::Shader::Sampler& entry) {
|
const VideoCommon::Shader::Sampler& entry) {
|
||||||
return TextureTypeToSurfaceTarget(entry.GetType(), entry.IsArray());
|
return TextureTypeToSurfaceTarget(entry.type, entry.is_array);
|
||||||
}
|
}
|
||||||
|
|
||||||
VideoCore::Surface::SurfaceTarget SurfaceParams::ExpectedTarget(
|
VideoCore::Surface::SurfaceTarget SurfaceParams::ExpectedTarget(
|
||||||
const VideoCommon::Shader::Image& entry) {
|
const VideoCommon::Shader::Image& entry) {
|
||||||
return ImageTypeToSurfaceTarget(entry.GetType());
|
return ImageTypeToSurfaceTarget(entry.type);
|
||||||
}
|
}
|
||||||
|
|
||||||
bool SurfaceParams::IsLayered() const {
|
bool SurfaceParams::IsLayered() const {
|
||||||
|
|
|
@ -1172,7 +1172,7 @@ private:
|
||||||
/// Returns true the shader sampler entry is compatible with the TIC texture type.
|
/// Returns true the shader sampler entry is compatible with the TIC texture type.
|
||||||
static bool IsTypeCompatible(Tegra::Texture::TextureType tic_type,
|
static bool IsTypeCompatible(Tegra::Texture::TextureType tic_type,
|
||||||
const VideoCommon::Shader::Sampler& entry) {
|
const VideoCommon::Shader::Sampler& entry) {
|
||||||
const auto shader_type = entry.GetType();
|
const auto shader_type = entry.type;
|
||||||
switch (tic_type) {
|
switch (tic_type) {
|
||||||
case Tegra::Texture::TextureType::Texture1D:
|
case Tegra::Texture::TextureType::Texture1D:
|
||||||
case Tegra::Texture::TextureType::Texture1DArray:
|
case Tegra::Texture::TextureType::Texture1DArray:
|
||||||
|
@ -1193,7 +1193,7 @@ private:
|
||||||
if (shader_type == Tegra::Shader::TextureType::TextureCube) {
|
if (shader_type == Tegra::Shader::TextureType::TextureCube) {
|
||||||
return true;
|
return true;
|
||||||
}
|
}
|
||||||
return shader_type == Tegra::Shader::TextureType::Texture2D && entry.IsArray();
|
return shader_type == Tegra::Shader::TextureType::Texture2D && entry.is_array;
|
||||||
}
|
}
|
||||||
UNREACHABLE();
|
UNREACHABLE();
|
||||||
return true;
|
return true;
|
||||||
|
|
Loading…
Reference in a new issue